diff --git "a/trainer_state.json" "b/trainer_state.json" new file mode 100644--- /dev/null +++ "b/trainer_state.json" @@ -0,0 +1,360019 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 3.645532703466294, + "eval_steps": 500, + "global_step": 60000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.723, + "step": 1 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.6266, + "step": 2 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.5433, + "step": 3 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.549, + "step": 4 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.4392, + "step": 5 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.4103, + "step": 6 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.3646, + "step": 7 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.406, + "step": 8 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.4313, + "step": 9 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.4064, + "step": 10 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.4813, + "step": 11 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.4032, + "step": 12 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.4155, + "step": 13 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.3646, + "step": 14 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.3787, + "step": 15 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.4208, + "step": 16 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.4073, + "step": 17 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.3681, + "step": 18 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.2523, + "step": 19 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.3873, + "step": 20 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.4356, + "step": 21 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.4044, + "step": 22 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.3472, + "step": 23 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.4316, + "step": 24 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.3745, + "step": 25 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.4373, + "step": 26 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.3434, + "step": 27 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.3634, + "step": 28 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.383, + "step": 29 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.3802, + "step": 30 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.3311, + "step": 31 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.2947, + "step": 32 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.3393, + "step": 33 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.3971, + "step": 34 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.3853, + "step": 35 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.3479, + "step": 36 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.327, + "step": 37 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.3567, + "step": 38 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.2774, + "step": 39 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.3087, + "step": 40 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.3209, + "step": 41 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.2982, + "step": 42 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.3023, + "step": 43 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.3805, + "step": 44 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.3128, + "step": 45 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.3346, + "step": 46 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.3723, + "step": 47 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.3082, + "step": 48 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.3435, + "step": 49 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.3326, + "step": 50 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.4064, + "step": 51 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.3746, + "step": 52 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.3569, + "step": 53 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.2699, + "step": 54 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.2927, + "step": 55 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.4283, + "step": 56 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.3213, + "step": 57 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.3252, + "step": 58 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.2934, + "step": 59 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.4053, + "step": 60 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.3496, + "step": 61 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.3098, + "step": 62 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.3207, + "step": 63 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.343, + "step": 64 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.34, + "step": 65 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.3325, + "step": 66 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.314, + "step": 67 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.3549, + "step": 68 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.3873, + "step": 69 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.3035, + "step": 70 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.3489, + "step": 71 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.3422, + "step": 72 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.3429, + "step": 73 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.2971, + "step": 74 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.336, + "step": 75 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.3102, + "step": 76 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.3285, + "step": 77 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.3526, + "step": 78 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.2926, + "step": 79 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.3176, + "step": 80 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.307, + "step": 81 + }, + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.2752, + "step": 82 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.315, + "step": 83 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2837, + "step": 84 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3007, + "step": 85 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3367, + "step": 86 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2945, + "step": 87 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.317, + "step": 88 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3104, + "step": 89 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3021, + "step": 90 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2852, + "step": 91 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3369, + "step": 92 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2431, + "step": 93 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2902, + "step": 94 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3311, + "step": 95 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3405, + "step": 96 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3078, + "step": 97 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.288, + "step": 98 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2792, + "step": 99 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3229, + "step": 100 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2788, + "step": 101 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3471, + "step": 102 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3148, + "step": 103 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3036, + "step": 104 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3281, + "step": 105 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3487, + "step": 106 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3192, + "step": 107 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3107, + "step": 108 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3137, + "step": 109 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.297, + "step": 110 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2855, + "step": 111 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2927, + "step": 112 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3396, + "step": 113 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3063, + "step": 114 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3521, + "step": 115 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2958, + "step": 116 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3641, + "step": 117 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2711, + "step": 118 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3022, + "step": 119 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2933, + "step": 120 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3269, + "step": 121 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2775, + "step": 122 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3091, + "step": 123 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2677, + "step": 124 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2997, + "step": 125 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2494, + "step": 126 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2947, + "step": 127 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2568, + "step": 128 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.4238, + "step": 129 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2809, + "step": 130 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2537, + "step": 131 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2728, + "step": 132 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3147, + "step": 133 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2749, + "step": 134 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2956, + "step": 135 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3259, + "step": 136 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3017, + "step": 137 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2872, + "step": 138 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3262, + "step": 139 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2812, + "step": 140 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2583, + "step": 141 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3105, + "step": 142 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2993, + "step": 143 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2793, + "step": 144 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3649, + "step": 145 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3747, + "step": 146 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3158, + "step": 147 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2484, + "step": 148 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2964, + "step": 149 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3054, + "step": 150 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.33, + "step": 151 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3155, + "step": 152 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2695, + "step": 153 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2291, + "step": 154 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2874, + "step": 155 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2892, + "step": 156 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3399, + "step": 157 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.275, + "step": 158 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.295, + "step": 159 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2329, + "step": 160 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3015, + "step": 161 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2844, + "step": 162 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3269, + "step": 163 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3134, + "step": 164 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3071, + "step": 165 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3546, + "step": 166 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3484, + "step": 167 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2113, + "step": 168 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2458, + "step": 169 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2493, + "step": 170 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.226, + "step": 171 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2681, + "step": 172 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2904, + "step": 173 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2317, + "step": 174 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3406, + "step": 175 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2967, + "step": 176 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2718, + "step": 177 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3227, + "step": 178 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2445, + "step": 179 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2268, + "step": 180 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2975, + "step": 181 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3252, + "step": 182 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2429, + "step": 183 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3145, + "step": 184 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2437, + "step": 185 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.243, + "step": 186 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2741, + "step": 187 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2706, + "step": 188 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2929, + "step": 189 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2379, + "step": 190 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3294, + "step": 191 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2589, + "step": 192 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3355, + "step": 193 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3497, + "step": 194 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3391, + "step": 195 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2602, + "step": 196 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2897, + "step": 197 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2601, + "step": 198 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2629, + "step": 199 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2549, + "step": 200 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2517, + "step": 201 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2357, + "step": 202 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2779, + "step": 203 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3035, + "step": 204 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3545, + "step": 205 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2402, + "step": 206 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2756, + "step": 207 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2851, + "step": 208 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2909, + "step": 209 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3018, + "step": 210 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3033, + "step": 211 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3363, + "step": 212 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3083, + "step": 213 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2369, + "step": 214 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2972, + "step": 215 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.352, + "step": 216 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2533, + "step": 217 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3139, + "step": 218 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2753, + "step": 219 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2482, + "step": 220 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3176, + "step": 221 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2394, + "step": 222 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2298, + "step": 223 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.265, + "step": 224 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2899, + "step": 225 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2516, + "step": 226 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2507, + "step": 227 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.264, + "step": 228 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3441, + "step": 229 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3098, + "step": 230 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2564, + "step": 231 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3248, + "step": 232 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2498, + "step": 233 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2964, + "step": 234 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2905, + "step": 235 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3331, + "step": 236 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2312, + "step": 237 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2919, + "step": 238 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2658, + "step": 239 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2568, + "step": 240 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3555, + "step": 241 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3145, + "step": 242 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3149, + "step": 243 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.3207, + "step": 244 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2912, + "step": 245 + }, + { + "epoch": 0.01, + "learning_rate": 2e-05, + "loss": 0.2321, + "step": 246 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2731, + "step": 247 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.3035, + "step": 248 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2317, + "step": 249 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2451, + "step": 250 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2367, + "step": 251 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2882, + "step": 252 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.276, + "step": 253 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.3185, + "step": 254 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2583, + "step": 255 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2644, + "step": 256 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2566, + "step": 257 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2548, + "step": 258 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2516, + "step": 259 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2313, + "step": 260 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.266, + "step": 261 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2491, + "step": 262 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2452, + "step": 263 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.315, + "step": 264 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2958, + "step": 265 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2541, + "step": 266 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2081, + "step": 267 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2803, + "step": 268 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2995, + "step": 269 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.3371, + "step": 270 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2865, + "step": 271 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2991, + "step": 272 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.3638, + "step": 273 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.3159, + "step": 274 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2819, + "step": 275 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2911, + "step": 276 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2719, + "step": 277 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.3028, + "step": 278 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2254, + "step": 279 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.3679, + "step": 280 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2552, + "step": 281 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2657, + "step": 282 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2407, + "step": 283 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.3313, + "step": 284 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2765, + "step": 285 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2612, + "step": 286 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.3313, + "step": 287 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2281, + "step": 288 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2982, + "step": 289 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2471, + "step": 290 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2536, + "step": 291 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2753, + "step": 292 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2548, + "step": 293 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.3259, + "step": 294 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2303, + "step": 295 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2802, + "step": 296 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2754, + "step": 297 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2806, + "step": 298 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2919, + "step": 299 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2608, + "step": 300 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.3069, + "step": 301 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2606, + "step": 302 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.312, + "step": 303 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2607, + "step": 304 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2964, + "step": 305 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2679, + "step": 306 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2776, + "step": 307 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2851, + "step": 308 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2757, + "step": 309 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2615, + "step": 310 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.3056, + "step": 311 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2272, + "step": 312 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2666, + "step": 313 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2399, + "step": 314 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.3583, + "step": 315 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.3334, + "step": 316 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2521, + "step": 317 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.3064, + "step": 318 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2203, + "step": 319 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2692, + "step": 320 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2514, + "step": 321 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2667, + "step": 322 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.294, + "step": 323 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2607, + "step": 324 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.3209, + "step": 325 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.231, + "step": 326 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.3198, + "step": 327 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.1768, + "step": 328 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.315, + "step": 329 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.3192, + "step": 330 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2525, + "step": 331 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2691, + "step": 332 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2977, + "step": 333 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2415, + "step": 334 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.3, + "step": 335 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2603, + "step": 336 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2771, + "step": 337 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2599, + "step": 338 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2814, + "step": 339 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2121, + "step": 340 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.3221, + "step": 341 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2488, + "step": 342 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2825, + "step": 343 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.208, + "step": 344 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2575, + "step": 345 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2982, + "step": 346 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2898, + "step": 347 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2837, + "step": 348 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2265, + "step": 349 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2751, + "step": 350 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.35, + "step": 351 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.3247, + "step": 352 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2044, + "step": 353 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.336, + "step": 354 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.3056, + "step": 355 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2511, + "step": 356 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2303, + "step": 357 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2712, + "step": 358 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2835, + "step": 359 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2809, + "step": 360 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.3271, + "step": 361 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2754, + "step": 362 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2364, + "step": 363 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.3058, + "step": 364 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.3347, + "step": 365 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2694, + "step": 366 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2913, + "step": 367 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2557, + "step": 368 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2179, + "step": 369 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2689, + "step": 370 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2103, + "step": 371 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2446, + "step": 372 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.247, + "step": 373 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2717, + "step": 374 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2085, + "step": 375 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2347, + "step": 376 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2639, + "step": 377 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.3288, + "step": 378 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.316, + "step": 379 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2711, + "step": 380 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2846, + "step": 381 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.3052, + "step": 382 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2877, + "step": 383 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2937, + "step": 384 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2998, + "step": 385 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2356, + "step": 386 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2787, + "step": 387 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2795, + "step": 388 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2372, + "step": 389 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2604, + "step": 390 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.3223, + "step": 391 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2948, + "step": 392 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2947, + "step": 393 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.298, + "step": 394 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2567, + "step": 395 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2752, + "step": 396 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2213, + "step": 397 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2874, + "step": 398 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2552, + "step": 399 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2542, + "step": 400 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2273, + "step": 401 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2703, + "step": 402 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2773, + "step": 403 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2893, + "step": 404 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.272, + "step": 405 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2931, + "step": 406 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2405, + "step": 407 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2279, + "step": 408 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2609, + "step": 409 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2809, + "step": 410 + }, + { + "epoch": 0.02, + "learning_rate": 2e-05, + "loss": 0.2351, + "step": 411 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2931, + "step": 412 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2434, + "step": 413 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2485, + "step": 414 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2834, + "step": 415 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2997, + "step": 416 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.1878, + "step": 417 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2597, + "step": 418 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2311, + "step": 419 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2738, + "step": 420 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2487, + "step": 421 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2738, + "step": 422 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2097, + "step": 423 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2431, + "step": 424 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.3533, + "step": 425 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2814, + "step": 426 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2277, + "step": 427 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2761, + "step": 428 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2624, + "step": 429 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2753, + "step": 430 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2314, + "step": 431 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2512, + "step": 432 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.263, + "step": 433 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2757, + "step": 434 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2252, + "step": 435 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.1875, + "step": 436 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2488, + "step": 437 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2598, + "step": 438 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2738, + "step": 439 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2638, + "step": 440 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.257, + "step": 441 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.3358, + "step": 442 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2763, + "step": 443 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2709, + "step": 444 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.263, + "step": 445 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2279, + "step": 446 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2191, + "step": 447 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2921, + "step": 448 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2511, + "step": 449 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2958, + "step": 450 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2861, + "step": 451 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2616, + "step": 452 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2489, + "step": 453 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.3032, + "step": 454 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.3235, + "step": 455 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2387, + "step": 456 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2693, + "step": 457 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2291, + "step": 458 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2593, + "step": 459 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2243, + "step": 460 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2426, + "step": 461 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2239, + "step": 462 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2596, + "step": 463 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.247, + "step": 464 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2704, + "step": 465 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2348, + "step": 466 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2347, + "step": 467 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.3193, + "step": 468 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.1991, + "step": 469 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2483, + "step": 470 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2678, + "step": 471 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2408, + "step": 472 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2999, + "step": 473 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2739, + "step": 474 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2577, + "step": 475 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2598, + "step": 476 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2381, + "step": 477 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2243, + "step": 478 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2948, + "step": 479 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.303, + "step": 480 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2745, + "step": 481 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2832, + "step": 482 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2463, + "step": 483 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2484, + "step": 484 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2284, + "step": 485 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2836, + "step": 486 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2916, + "step": 487 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2312, + "step": 488 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2402, + "step": 489 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2235, + "step": 490 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2905, + "step": 491 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2591, + "step": 492 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2505, + "step": 493 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2094, + "step": 494 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.256, + "step": 495 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2604, + "step": 496 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2233, + "step": 497 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.1933, + "step": 498 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2506, + "step": 499 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2206, + "step": 500 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2007, + "step": 501 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.1901, + "step": 502 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.1962, + "step": 503 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2505, + "step": 504 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2396, + "step": 505 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2911, + "step": 506 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.1878, + "step": 507 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2256, + "step": 508 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2804, + "step": 509 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2952, + "step": 510 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.206, + "step": 511 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2359, + "step": 512 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.268, + "step": 513 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2411, + "step": 514 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2785, + "step": 515 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.275, + "step": 516 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2344, + "step": 517 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2601, + "step": 518 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.221, + "step": 519 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2699, + "step": 520 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2689, + "step": 521 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.259, + "step": 522 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.288, + "step": 523 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.1989, + "step": 524 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2519, + "step": 525 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2839, + "step": 526 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2424, + "step": 527 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.286, + "step": 528 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2456, + "step": 529 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2629, + "step": 530 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2229, + "step": 531 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2408, + "step": 532 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2357, + "step": 533 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2612, + "step": 534 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2232, + "step": 535 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2502, + "step": 536 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2742, + "step": 537 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2969, + "step": 538 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2722, + "step": 539 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2718, + "step": 540 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2624, + "step": 541 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2208, + "step": 542 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2244, + "step": 543 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2243, + "step": 544 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2425, + "step": 545 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2993, + "step": 546 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2944, + "step": 547 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2233, + "step": 548 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2031, + "step": 549 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2058, + "step": 550 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2873, + "step": 551 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2579, + "step": 552 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2618, + "step": 553 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2982, + "step": 554 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2953, + "step": 555 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.208, + "step": 556 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2687, + "step": 557 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2678, + "step": 558 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.3046, + "step": 559 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2812, + "step": 560 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2933, + "step": 561 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2727, + "step": 562 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2342, + "step": 563 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.272, + "step": 564 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2336, + "step": 565 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2607, + "step": 566 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.3013, + "step": 567 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2387, + "step": 568 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2811, + "step": 569 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2575, + "step": 570 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2214, + "step": 571 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2664, + "step": 572 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2791, + "step": 573 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2748, + "step": 574 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.2338, + "step": 575 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.3283, + "step": 576 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2562, + "step": 577 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2783, + "step": 578 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2513, + "step": 579 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2723, + "step": 580 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2355, + "step": 581 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2888, + "step": 582 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2562, + "step": 583 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2225, + "step": 584 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2339, + "step": 585 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.23, + "step": 586 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2548, + "step": 587 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2386, + "step": 588 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2416, + "step": 589 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2223, + "step": 590 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2533, + "step": 591 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.3195, + "step": 592 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2939, + "step": 593 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2756, + "step": 594 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 595 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2757, + "step": 596 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2836, + "step": 597 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2475, + "step": 598 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2785, + "step": 599 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2682, + "step": 600 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.216, + "step": 601 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2723, + "step": 602 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2568, + "step": 603 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2432, + "step": 604 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.273, + "step": 605 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.247, + "step": 606 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2888, + "step": 607 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.255, + "step": 608 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2852, + "step": 609 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2597, + "step": 610 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2291, + "step": 611 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.194, + "step": 612 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2889, + "step": 613 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2746, + "step": 614 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2841, + "step": 615 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.1987, + "step": 616 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.216, + "step": 617 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2015, + "step": 618 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2751, + "step": 619 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.1914, + "step": 620 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2139, + "step": 621 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2287, + "step": 622 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2744, + "step": 623 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2417, + "step": 624 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.3196, + "step": 625 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.23, + "step": 626 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2619, + "step": 627 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2226, + "step": 628 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2567, + "step": 629 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2773, + "step": 630 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2672, + "step": 631 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2565, + "step": 632 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2071, + "step": 633 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2283, + "step": 634 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2088, + "step": 635 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2786, + "step": 636 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2239, + "step": 637 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2261, + "step": 638 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.259, + "step": 639 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2483, + "step": 640 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2669, + "step": 641 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2283, + "step": 642 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2591, + "step": 643 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2019, + "step": 644 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2756, + "step": 645 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.258, + "step": 646 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.3277, + "step": 647 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.211, + "step": 648 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2054, + "step": 649 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.244, + "step": 650 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2213, + "step": 651 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2423, + "step": 652 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2481, + "step": 653 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.1989, + "step": 654 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.1932, + "step": 655 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2296, + "step": 656 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2474, + "step": 657 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2195, + "step": 658 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2637, + "step": 659 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2256, + "step": 660 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2046, + "step": 661 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2496, + "step": 662 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2623, + "step": 663 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2204, + "step": 664 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2529, + "step": 665 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2374, + "step": 666 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2574, + "step": 667 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2646, + "step": 668 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2371, + "step": 669 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2309, + "step": 670 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.1935, + "step": 671 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.1978, + "step": 672 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2559, + "step": 673 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.3037, + "step": 674 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2534, + "step": 675 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2262, + "step": 676 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2769, + "step": 677 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2674, + "step": 678 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2031, + "step": 679 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2655, + "step": 680 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2183, + "step": 681 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2268, + "step": 682 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2437, + "step": 683 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2821, + "step": 684 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2393, + "step": 685 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2093, + "step": 686 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.219, + "step": 687 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2383, + "step": 688 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2493, + "step": 689 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2406, + "step": 690 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2538, + "step": 691 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2625, + "step": 692 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2425, + "step": 693 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2243, + "step": 694 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2323, + "step": 695 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2663, + "step": 696 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.3015, + "step": 697 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.3345, + "step": 698 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2786, + "step": 699 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2559, + "step": 700 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2152, + "step": 701 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2647, + "step": 702 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2499, + "step": 703 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.1925, + "step": 704 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2856, + "step": 705 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2932, + "step": 706 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2454, + "step": 707 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2536, + "step": 708 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2137, + "step": 709 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.3114, + "step": 710 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2366, + "step": 711 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2621, + "step": 712 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2296, + "step": 713 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2721, + "step": 714 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2503, + "step": 715 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2245, + "step": 716 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2511, + "step": 717 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.3314, + "step": 718 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.261, + "step": 719 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2505, + "step": 720 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2518, + "step": 721 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2593, + "step": 722 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.3616, + "step": 723 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2785, + "step": 724 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2484, + "step": 725 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2425, + "step": 726 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2527, + "step": 727 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2464, + "step": 728 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2693, + "step": 729 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2518, + "step": 730 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.25, + "step": 731 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2607, + "step": 732 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2209, + "step": 733 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2776, + "step": 734 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2729, + "step": 735 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2656, + "step": 736 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2259, + "step": 737 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2029, + "step": 738 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2207, + "step": 739 + }, + { + "epoch": 0.04, + "learning_rate": 2e-05, + "loss": 0.2607, + "step": 740 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2264, + "step": 741 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.255, + "step": 742 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2225, + "step": 743 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2441, + "step": 744 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2616, + "step": 745 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2204, + "step": 746 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2534, + "step": 747 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2941, + "step": 748 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2728, + "step": 749 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.3109, + "step": 750 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2014, + "step": 751 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.299, + "step": 752 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2574, + "step": 753 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2229, + "step": 754 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.3015, + "step": 755 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.253, + "step": 756 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2536, + "step": 757 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2328, + "step": 758 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2549, + "step": 759 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2681, + "step": 760 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2625, + "step": 761 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.233, + "step": 762 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2579, + "step": 763 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2198, + "step": 764 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.22, + "step": 765 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.3145, + "step": 766 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2343, + "step": 767 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.1761, + "step": 768 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2142, + "step": 769 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2848, + "step": 770 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2629, + "step": 771 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.222, + "step": 772 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2192, + "step": 773 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2529, + "step": 774 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2597, + "step": 775 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.239, + "step": 776 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2484, + "step": 777 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.219, + "step": 778 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2127, + "step": 779 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2043, + "step": 780 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.252, + "step": 781 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2644, + "step": 782 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2261, + "step": 783 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2427, + "step": 784 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2393, + "step": 785 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2732, + "step": 786 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2473, + "step": 787 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2221, + "step": 788 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2098, + "step": 789 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2457, + "step": 790 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2385, + "step": 791 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2645, + "step": 792 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2064, + "step": 793 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2329, + "step": 794 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2077, + "step": 795 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2319, + "step": 796 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2536, + "step": 797 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.232, + "step": 798 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2788, + "step": 799 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2652, + "step": 800 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.1742, + "step": 801 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2056, + "step": 802 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2555, + "step": 803 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.249, + "step": 804 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2624, + "step": 805 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2864, + "step": 806 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2117, + "step": 807 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.1861, + "step": 808 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2199, + "step": 809 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2687, + "step": 810 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2627, + "step": 811 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.3372, + "step": 812 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2024, + "step": 813 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2127, + "step": 814 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.1634, + "step": 815 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2222, + "step": 816 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.1993, + "step": 817 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2466, + "step": 818 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.21, + "step": 819 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2658, + "step": 820 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.202, + "step": 821 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2374, + "step": 822 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2498, + "step": 823 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2515, + "step": 824 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2552, + "step": 825 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.1976, + "step": 826 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2366, + "step": 827 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2328, + "step": 828 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2456, + "step": 829 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2226, + "step": 830 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2247, + "step": 831 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2148, + "step": 832 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.1968, + "step": 833 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.3279, + "step": 834 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2397, + "step": 835 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2358, + "step": 836 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.1802, + "step": 837 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2531, + "step": 838 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2349, + "step": 839 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2173, + "step": 840 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2852, + "step": 841 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2527, + "step": 842 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2679, + "step": 843 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2786, + "step": 844 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.223, + "step": 845 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2825, + "step": 846 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2653, + "step": 847 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2409, + "step": 848 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.211, + "step": 849 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2285, + "step": 850 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2363, + "step": 851 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.3247, + "step": 852 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2511, + "step": 853 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2386, + "step": 854 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2712, + "step": 855 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2243, + "step": 856 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2582, + "step": 857 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2393, + "step": 858 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2494, + "step": 859 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2715, + "step": 860 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2263, + "step": 861 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2259, + "step": 862 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2555, + "step": 863 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2508, + "step": 864 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2738, + "step": 865 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2384, + "step": 866 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2491, + "step": 867 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2773, + "step": 868 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2421, + "step": 869 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2344, + "step": 870 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.1813, + "step": 871 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2645, + "step": 872 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 873 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2516, + "step": 874 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2482, + "step": 875 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2855, + "step": 876 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2038, + "step": 877 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.237, + "step": 878 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2098, + "step": 879 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2351, + "step": 880 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.1912, + "step": 881 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2328, + "step": 882 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2627, + "step": 883 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2106, + "step": 884 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2262, + "step": 885 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2257, + "step": 886 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2315, + "step": 887 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2296, + "step": 888 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2367, + "step": 889 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2899, + "step": 890 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2519, + "step": 891 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2456, + "step": 892 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2654, + "step": 893 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2562, + "step": 894 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2579, + "step": 895 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.3169, + "step": 896 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2388, + "step": 897 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2039, + "step": 898 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2177, + "step": 899 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2492, + "step": 900 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2834, + "step": 901 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.1982, + "step": 902 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2359, + "step": 903 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2359, + "step": 904 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "loss": 0.2582, + "step": 905 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.261, + "step": 906 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2445, + "step": 907 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2713, + "step": 908 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2641, + "step": 909 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2235, + "step": 910 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2241, + "step": 911 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2466, + "step": 912 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2105, + "step": 913 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2492, + "step": 914 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2411, + "step": 915 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.1853, + "step": 916 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.1976, + "step": 917 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2779, + "step": 918 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2502, + "step": 919 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.274, + "step": 920 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2489, + "step": 921 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.1771, + "step": 922 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2213, + "step": 923 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.3187, + "step": 924 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2238, + "step": 925 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.288, + "step": 926 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.1713, + "step": 927 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2643, + "step": 928 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.1817, + "step": 929 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2008, + "step": 930 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2511, + "step": 931 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2308, + "step": 932 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2447, + "step": 933 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.237, + "step": 934 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2737, + "step": 935 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2393, + "step": 936 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2129, + "step": 937 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2634, + "step": 938 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2621, + "step": 939 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2878, + "step": 940 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2585, + "step": 941 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2377, + "step": 942 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2724, + "step": 943 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.3027, + "step": 944 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2252, + "step": 945 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2289, + "step": 946 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2176, + "step": 947 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2831, + "step": 948 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2094, + "step": 949 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2367, + "step": 950 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.275, + "step": 951 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2461, + "step": 952 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2824, + "step": 953 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2389, + "step": 954 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.1969, + "step": 955 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.1796, + "step": 956 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2657, + "step": 957 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2555, + "step": 958 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.1981, + "step": 959 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2239, + "step": 960 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.1817, + "step": 961 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2217, + "step": 962 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.209, + "step": 963 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2468, + "step": 964 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2448, + "step": 965 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2912, + "step": 966 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2237, + "step": 967 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.1767, + "step": 968 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2327, + "step": 969 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.264, + "step": 970 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2291, + "step": 971 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2597, + "step": 972 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.1839, + "step": 973 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.1982, + "step": 974 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.1694, + "step": 975 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2187, + "step": 976 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2271, + "step": 977 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2316, + "step": 978 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2057, + "step": 979 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.3236, + "step": 980 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2587, + "step": 981 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2148, + "step": 982 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.277, + "step": 983 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2271, + "step": 984 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2953, + "step": 985 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2269, + "step": 986 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2689, + "step": 987 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2607, + "step": 988 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2677, + "step": 989 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.1761, + "step": 990 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2646, + "step": 991 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 992 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2567, + "step": 993 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.1925, + "step": 994 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2291, + "step": 995 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2457, + "step": 996 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2468, + "step": 997 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.286, + "step": 998 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2437, + "step": 999 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2313, + "step": 1000 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2643, + "step": 1001 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2686, + "step": 1002 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2129, + "step": 1003 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.27, + "step": 1004 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2171, + "step": 1005 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.1973, + "step": 1006 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2373, + "step": 1007 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2124, + "step": 1008 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.1936, + "step": 1009 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2227, + "step": 1010 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2113, + "step": 1011 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2434, + "step": 1012 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2819, + "step": 1013 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2893, + "step": 1014 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2185, + "step": 1015 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2452, + "step": 1016 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2582, + "step": 1017 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2555, + "step": 1018 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.1908, + "step": 1019 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2732, + "step": 1020 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2353, + "step": 1021 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2794, + "step": 1022 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2436, + "step": 1023 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2577, + "step": 1024 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2467, + "step": 1025 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2867, + "step": 1026 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2192, + "step": 1027 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2664, + "step": 1028 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2001, + "step": 1029 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.1827, + "step": 1030 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2432, + "step": 1031 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2809, + "step": 1032 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.1934, + "step": 1033 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2638, + "step": 1034 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.1685, + "step": 1035 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.1932, + "step": 1036 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2645, + "step": 1037 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2527, + "step": 1038 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2487, + "step": 1039 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2781, + "step": 1040 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2291, + "step": 1041 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2337, + "step": 1042 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2231, + "step": 1043 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2393, + "step": 1044 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.1611, + "step": 1045 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2243, + "step": 1046 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2335, + "step": 1047 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.214, + "step": 1048 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 1049 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2551, + "step": 1050 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2274, + "step": 1051 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.188, + "step": 1052 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2335, + "step": 1053 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2374, + "step": 1054 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2731, + "step": 1055 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2201, + "step": 1056 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.232, + "step": 1057 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2861, + "step": 1058 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2529, + "step": 1059 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.3021, + "step": 1060 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2218, + "step": 1061 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2204, + "step": 1062 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.27, + "step": 1063 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2135, + "step": 1064 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2621, + "step": 1065 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2451, + "step": 1066 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.1878, + "step": 1067 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2421, + "step": 1068 + }, + { + "epoch": 0.06, + "learning_rate": 2e-05, + "loss": 0.2638, + "step": 1069 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2685, + "step": 1070 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2301, + "step": 1071 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.3045, + "step": 1072 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2236, + "step": 1073 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2147, + "step": 1074 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.268, + "step": 1075 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2538, + "step": 1076 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.1864, + "step": 1077 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2365, + "step": 1078 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2257, + "step": 1079 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.1993, + "step": 1080 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2185, + "step": 1081 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2742, + "step": 1082 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2361, + "step": 1083 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2266, + "step": 1084 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2939, + "step": 1085 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2281, + "step": 1086 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2336, + "step": 1087 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2048, + "step": 1088 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2133, + "step": 1089 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2072, + "step": 1090 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2066, + "step": 1091 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.1687, + "step": 1092 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2159, + "step": 1093 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2026, + "step": 1094 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.238, + "step": 1095 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.1794, + "step": 1096 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2041, + "step": 1097 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.239, + "step": 1098 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2309, + "step": 1099 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2699, + "step": 1100 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2419, + "step": 1101 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.1938, + "step": 1102 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.1966, + "step": 1103 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2343, + "step": 1104 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.243, + "step": 1105 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2316, + "step": 1106 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2396, + "step": 1107 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2545, + "step": 1108 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.264, + "step": 1109 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2628, + "step": 1110 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.253, + "step": 1111 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2568, + "step": 1112 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.201, + "step": 1113 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.3353, + "step": 1114 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2285, + "step": 1115 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2399, + "step": 1116 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2443, + "step": 1117 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.1588, + "step": 1118 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2217, + "step": 1119 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2471, + "step": 1120 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2146, + "step": 1121 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2823, + "step": 1122 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.3017, + "step": 1123 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2087, + "step": 1124 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2389, + "step": 1125 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2322, + "step": 1126 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2214, + "step": 1127 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.308, + "step": 1128 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2362, + "step": 1129 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2343, + "step": 1130 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.205, + "step": 1131 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2435, + "step": 1132 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2442, + "step": 1133 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2184, + "step": 1134 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.1889, + "step": 1135 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.1991, + "step": 1136 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2319, + "step": 1137 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2394, + "step": 1138 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.3029, + "step": 1139 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.1882, + "step": 1140 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.181, + "step": 1141 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2669, + "step": 1142 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2186, + "step": 1143 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2238, + "step": 1144 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2047, + "step": 1145 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2719, + "step": 1146 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.1933, + "step": 1147 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2458, + "step": 1148 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2513, + "step": 1149 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2324, + "step": 1150 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2125, + "step": 1151 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2158, + "step": 1152 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2135, + "step": 1153 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.1904, + "step": 1154 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2408, + "step": 1155 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.243, + "step": 1156 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2316, + "step": 1157 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.223, + "step": 1158 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2765, + "step": 1159 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.1841, + "step": 1160 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.1914, + "step": 1161 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2724, + "step": 1162 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.1964, + "step": 1163 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2277, + "step": 1164 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2528, + "step": 1165 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2353, + "step": 1166 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.1699, + "step": 1167 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2811, + "step": 1168 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.239, + "step": 1169 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.1793, + "step": 1170 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.241, + "step": 1171 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2357, + "step": 1172 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2683, + "step": 1173 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.1907, + "step": 1174 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2092, + "step": 1175 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2405, + "step": 1176 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2384, + "step": 1177 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.196, + "step": 1178 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2406, + "step": 1179 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2485, + "step": 1180 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.1485, + "step": 1181 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2612, + "step": 1182 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2093, + "step": 1183 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2651, + "step": 1184 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2349, + "step": 1185 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2108, + "step": 1186 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2324, + "step": 1187 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2343, + "step": 1188 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2822, + "step": 1189 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.3188, + "step": 1190 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2806, + "step": 1191 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.1947, + "step": 1192 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2382, + "step": 1193 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2596, + "step": 1194 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.21, + "step": 1195 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2278, + "step": 1196 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.1759, + "step": 1197 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.234, + "step": 1198 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2684, + "step": 1199 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2515, + "step": 1200 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2767, + "step": 1201 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2607, + "step": 1202 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2868, + "step": 1203 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.1791, + "step": 1204 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2219, + "step": 1205 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2089, + "step": 1206 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2653, + "step": 1207 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2381, + "step": 1208 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.214, + "step": 1209 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.225, + "step": 1210 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2289, + "step": 1211 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2591, + "step": 1212 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2242, + "step": 1213 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.3153, + "step": 1214 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2604, + "step": 1215 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.189, + "step": 1216 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2745, + "step": 1217 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2819, + "step": 1218 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2558, + "step": 1219 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.215, + "step": 1220 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.249, + "step": 1221 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.1939, + "step": 1222 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2209, + "step": 1223 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.1971, + "step": 1224 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2653, + "step": 1225 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.1757, + "step": 1226 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2207, + "step": 1227 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2478, + "step": 1228 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2514, + "step": 1229 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2387, + "step": 1230 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2306, + "step": 1231 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2363, + "step": 1232 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2267, + "step": 1233 + }, + { + "epoch": 0.07, + "learning_rate": 2e-05, + "loss": 0.2328, + "step": 1234 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2741, + "step": 1235 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.215, + "step": 1236 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2613, + "step": 1237 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2312, + "step": 1238 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2248, + "step": 1239 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2034, + "step": 1240 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 1241 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2141, + "step": 1242 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2339, + "step": 1243 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2132, + "step": 1244 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2751, + "step": 1245 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2629, + "step": 1246 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.289, + "step": 1247 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.1921, + "step": 1248 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2306, + "step": 1249 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2761, + "step": 1250 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2367, + "step": 1251 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2126, + "step": 1252 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2254, + "step": 1253 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.1958, + "step": 1254 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2569, + "step": 1255 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2704, + "step": 1256 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2206, + "step": 1257 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2162, + "step": 1258 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2397, + "step": 1259 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.207, + "step": 1260 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2564, + "step": 1261 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2499, + "step": 1262 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2086, + "step": 1263 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2047, + "step": 1264 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2662, + "step": 1265 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2269, + "step": 1266 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2555, + "step": 1267 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2695, + "step": 1268 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.1955, + "step": 1269 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.1859, + "step": 1270 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.234, + "step": 1271 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.1967, + "step": 1272 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2161, + "step": 1273 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2724, + "step": 1274 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2328, + "step": 1275 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2277, + "step": 1276 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2119, + "step": 1277 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.194, + "step": 1278 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2016, + "step": 1279 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2612, + "step": 1280 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2025, + "step": 1281 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2568, + "step": 1282 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2433, + "step": 1283 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.1808, + "step": 1284 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2547, + "step": 1285 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2562, + "step": 1286 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.191, + "step": 1287 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.232, + "step": 1288 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2108, + "step": 1289 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2357, + "step": 1290 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2372, + "step": 1291 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2452, + "step": 1292 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2229, + "step": 1293 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.1766, + "step": 1294 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2865, + "step": 1295 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2334, + "step": 1296 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2406, + "step": 1297 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2253, + "step": 1298 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2715, + "step": 1299 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2408, + "step": 1300 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.1956, + "step": 1301 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.1938, + "step": 1302 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2519, + "step": 1303 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2513, + "step": 1304 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2666, + "step": 1305 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2373, + "step": 1306 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.212, + "step": 1307 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2534, + "step": 1308 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2265, + "step": 1309 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2022, + "step": 1310 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.19, + "step": 1311 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.239, + "step": 1312 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2429, + "step": 1313 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2198, + "step": 1314 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2893, + "step": 1315 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2568, + "step": 1316 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2613, + "step": 1317 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2299, + "step": 1318 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2536, + "step": 1319 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.3033, + "step": 1320 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2531, + "step": 1321 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2251, + "step": 1322 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2034, + "step": 1323 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2184, + "step": 1324 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2508, + "step": 1325 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2782, + "step": 1326 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2304, + "step": 1327 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2794, + "step": 1328 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2685, + "step": 1329 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2246, + "step": 1330 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2418, + "step": 1331 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2652, + "step": 1332 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2291, + "step": 1333 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2595, + "step": 1334 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.303, + "step": 1335 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2583, + "step": 1336 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2398, + "step": 1337 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2356, + "step": 1338 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.1838, + "step": 1339 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.1707, + "step": 1340 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2025, + "step": 1341 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2458, + "step": 1342 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2115, + "step": 1343 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2457, + "step": 1344 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2022, + "step": 1345 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2432, + "step": 1346 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2715, + "step": 1347 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.1808, + "step": 1348 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2528, + "step": 1349 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2382, + "step": 1350 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2608, + "step": 1351 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2014, + "step": 1352 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2283, + "step": 1353 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.1955, + "step": 1354 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2144, + "step": 1355 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2477, + "step": 1356 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2077, + "step": 1357 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2108, + "step": 1358 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2615, + "step": 1359 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2105, + "step": 1360 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2477, + "step": 1361 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2011, + "step": 1362 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.1729, + "step": 1363 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2142, + "step": 1364 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2273, + "step": 1365 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2338, + "step": 1366 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.3194, + "step": 1367 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2197, + "step": 1368 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2398, + "step": 1369 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2238, + "step": 1370 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2663, + "step": 1371 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.221, + "step": 1372 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2118, + "step": 1373 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2627, + "step": 1374 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2091, + "step": 1375 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2289, + "step": 1376 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2267, + "step": 1377 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.1933, + "step": 1378 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2314, + "step": 1379 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2543, + "step": 1380 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.1987, + "step": 1381 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2197, + "step": 1382 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2067, + "step": 1383 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2676, + "step": 1384 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2174, + "step": 1385 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2147, + "step": 1386 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2296, + "step": 1387 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.1735, + "step": 1388 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.183, + "step": 1389 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.241, + "step": 1390 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2606, + "step": 1391 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2468, + "step": 1392 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2874, + "step": 1393 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2978, + "step": 1394 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2437, + "step": 1395 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2397, + "step": 1396 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2074, + "step": 1397 + }, + { + "epoch": 0.08, + "learning_rate": 2e-05, + "loss": 0.2861, + "step": 1398 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2185, + "step": 1399 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2194, + "step": 1400 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.278, + "step": 1401 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.1964, + "step": 1402 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.206, + "step": 1403 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2123, + "step": 1404 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2836, + "step": 1405 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.3041, + "step": 1406 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2183, + "step": 1407 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2629, + "step": 1408 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2773, + "step": 1409 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2489, + "step": 1410 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2534, + "step": 1411 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2436, + "step": 1412 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.1759, + "step": 1413 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2887, + "step": 1414 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2143, + "step": 1415 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2362, + "step": 1416 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2367, + "step": 1417 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2445, + "step": 1418 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2324, + "step": 1419 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2117, + "step": 1420 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.1839, + "step": 1421 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2737, + "step": 1422 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2079, + "step": 1423 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2551, + "step": 1424 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.187, + "step": 1425 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2478, + "step": 1426 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2164, + "step": 1427 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2085, + "step": 1428 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2447, + "step": 1429 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2183, + "step": 1430 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2825, + "step": 1431 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2501, + "step": 1432 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2092, + "step": 1433 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.1894, + "step": 1434 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2085, + "step": 1435 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.248, + "step": 1436 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.1646, + "step": 1437 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2282, + "step": 1438 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2182, + "step": 1439 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2326, + "step": 1440 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2149, + "step": 1441 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2132, + "step": 1442 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.1844, + "step": 1443 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2492, + "step": 1444 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2304, + "step": 1445 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.1975, + "step": 1446 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2214, + "step": 1447 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.24, + "step": 1448 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2703, + "step": 1449 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2431, + "step": 1450 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2135, + "step": 1451 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2046, + "step": 1452 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2642, + "step": 1453 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.222, + "step": 1454 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2149, + "step": 1455 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2884, + "step": 1456 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2282, + "step": 1457 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2778, + "step": 1458 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2302, + "step": 1459 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2463, + "step": 1460 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.1693, + "step": 1461 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2441, + "step": 1462 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.232, + "step": 1463 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2563, + "step": 1464 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2551, + "step": 1465 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2478, + "step": 1466 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.1856, + "step": 1467 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.259, + "step": 1468 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2198, + "step": 1469 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2769, + "step": 1470 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2616, + "step": 1471 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2153, + "step": 1472 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2625, + "step": 1473 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2166, + "step": 1474 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 1475 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2527, + "step": 1476 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2819, + "step": 1477 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.216, + "step": 1478 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2058, + "step": 1479 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2471, + "step": 1480 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.1998, + "step": 1481 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2205, + "step": 1482 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2374, + "step": 1483 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.1548, + "step": 1484 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2608, + "step": 1485 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.1746, + "step": 1486 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.231, + "step": 1487 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2747, + "step": 1488 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2414, + "step": 1489 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2348, + "step": 1490 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2541, + "step": 1491 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.1687, + "step": 1492 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2022, + "step": 1493 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.1882, + "step": 1494 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2226, + "step": 1495 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2241, + "step": 1496 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2342, + "step": 1497 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2719, + "step": 1498 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2383, + "step": 1499 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.1952, + "step": 1500 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2239, + "step": 1501 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2004, + "step": 1502 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2043, + "step": 1503 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.1957, + "step": 1504 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.231, + "step": 1505 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2426, + "step": 1506 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.1877, + "step": 1507 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.235, + "step": 1508 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2448, + "step": 1509 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2803, + "step": 1510 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.1775, + "step": 1511 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2519, + "step": 1512 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2025, + "step": 1513 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.1909, + "step": 1514 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2755, + "step": 1515 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2141, + "step": 1516 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.1941, + "step": 1517 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2199, + "step": 1518 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.1739, + "step": 1519 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.1905, + "step": 1520 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.221, + "step": 1521 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2145, + "step": 1522 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2226, + "step": 1523 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2155, + "step": 1524 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2149, + "step": 1525 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.1936, + "step": 1526 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.238, + "step": 1527 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2462, + "step": 1528 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2193, + "step": 1529 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2139, + "step": 1530 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2027, + "step": 1531 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2158, + "step": 1532 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2105, + "step": 1533 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.1896, + "step": 1534 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2458, + "step": 1535 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.1714, + "step": 1536 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2384, + "step": 1537 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2472, + "step": 1538 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2443, + "step": 1539 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2061, + "step": 1540 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.1909, + "step": 1541 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2327, + "step": 1542 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.19, + "step": 1543 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2194, + "step": 1544 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2182, + "step": 1545 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.3188, + "step": 1546 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2155, + "step": 1547 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2258, + "step": 1548 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.1902, + "step": 1549 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2209, + "step": 1550 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.1827, + "step": 1551 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2143, + "step": 1552 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.1842, + "step": 1553 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2426, + "step": 1554 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.261, + "step": 1555 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2233, + "step": 1556 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.1627, + "step": 1557 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2309, + "step": 1558 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2166, + "step": 1559 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2419, + "step": 1560 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2274, + "step": 1561 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2129, + "step": 1562 + }, + { + "epoch": 0.09, + "learning_rate": 2e-05, + "loss": 0.2208, + "step": 1563 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.1871, + "step": 1564 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2309, + "step": 1565 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.1635, + "step": 1566 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2235, + "step": 1567 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2291, + "step": 1568 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 1569 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2384, + "step": 1570 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2086, + "step": 1571 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2389, + "step": 1572 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.1722, + "step": 1573 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2165, + "step": 1574 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2428, + "step": 1575 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.24, + "step": 1576 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2046, + "step": 1577 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2539, + "step": 1578 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.1682, + "step": 1579 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2193, + "step": 1580 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 1581 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.1691, + "step": 1582 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2895, + "step": 1583 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2113, + "step": 1584 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2718, + "step": 1585 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2296, + "step": 1586 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2327, + "step": 1587 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2085, + "step": 1588 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2384, + "step": 1589 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.1985, + "step": 1590 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.1967, + "step": 1591 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2348, + "step": 1592 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2314, + "step": 1593 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.1829, + "step": 1594 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2417, + "step": 1595 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2219, + "step": 1596 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.1693, + "step": 1597 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.1753, + "step": 1598 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.3465, + "step": 1599 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2352, + "step": 1600 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.1801, + "step": 1601 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2092, + "step": 1602 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2412, + "step": 1603 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.1852, + "step": 1604 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2873, + "step": 1605 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2522, + "step": 1606 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2502, + "step": 1607 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2269, + "step": 1608 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2003, + "step": 1609 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2675, + "step": 1610 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2607, + "step": 1611 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2136, + "step": 1612 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.1898, + "step": 1613 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2408, + "step": 1614 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.153, + "step": 1615 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2511, + "step": 1616 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2649, + "step": 1617 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.1872, + "step": 1618 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2545, + "step": 1619 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2178, + "step": 1620 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2464, + "step": 1621 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2342, + "step": 1622 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2278, + "step": 1623 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2434, + "step": 1624 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.1824, + "step": 1625 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2078, + "step": 1626 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2396, + "step": 1627 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2372, + "step": 1628 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.1641, + "step": 1629 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2392, + "step": 1630 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2841, + "step": 1631 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.1865, + "step": 1632 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2282, + "step": 1633 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.235, + "step": 1634 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2725, + "step": 1635 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2008, + "step": 1636 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.3252, + "step": 1637 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2007, + "step": 1638 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2336, + "step": 1639 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.251, + "step": 1640 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2451, + "step": 1641 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2027, + "step": 1642 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2408, + "step": 1643 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 1644 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2279, + "step": 1645 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2725, + "step": 1646 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.1718, + "step": 1647 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2159, + "step": 1648 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2081, + "step": 1649 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2204, + "step": 1650 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.268, + "step": 1651 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2391, + "step": 1652 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2442, + "step": 1653 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2362, + "step": 1654 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.1679, + "step": 1655 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2359, + "step": 1656 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.262, + "step": 1657 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.1839, + "step": 1658 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2198, + "step": 1659 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2593, + "step": 1660 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2224, + "step": 1661 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2523, + "step": 1662 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.1902, + "step": 1663 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2318, + "step": 1664 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2821, + "step": 1665 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2238, + "step": 1666 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.1741, + "step": 1667 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2043, + "step": 1668 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2199, + "step": 1669 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.1881, + "step": 1670 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2634, + "step": 1671 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2222, + "step": 1672 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2367, + "step": 1673 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2642, + "step": 1674 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2551, + "step": 1675 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2362, + "step": 1676 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2156, + "step": 1677 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2406, + "step": 1678 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2038, + "step": 1679 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.243, + "step": 1680 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.1998, + "step": 1681 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2237, + "step": 1682 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.1858, + "step": 1683 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2311, + "step": 1684 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.208, + "step": 1685 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2176, + "step": 1686 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2218, + "step": 1687 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.155, + "step": 1688 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2527, + "step": 1689 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.214, + "step": 1690 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2124, + "step": 1691 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2364, + "step": 1692 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.187, + "step": 1693 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.1948, + "step": 1694 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2532, + "step": 1695 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2569, + "step": 1696 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2019, + "step": 1697 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2293, + "step": 1698 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2044, + "step": 1699 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.1916, + "step": 1700 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.1945, + "step": 1701 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2271, + "step": 1702 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2485, + "step": 1703 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2333, + "step": 1704 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2173, + "step": 1705 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.1757, + "step": 1706 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2251, + "step": 1707 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2295, + "step": 1708 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.1934, + "step": 1709 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.1956, + "step": 1710 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2644, + "step": 1711 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2292, + "step": 1712 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.1995, + "step": 1713 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.1983, + "step": 1714 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2466, + "step": 1715 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.1705, + "step": 1716 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2478, + "step": 1717 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2106, + "step": 1718 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.1936, + "step": 1719 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.3236, + "step": 1720 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2685, + "step": 1721 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2009, + "step": 1722 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.1674, + "step": 1723 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2137, + "step": 1724 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2338, + "step": 1725 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2208, + "step": 1726 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.221, + "step": 1727 + }, + { + "epoch": 0.1, + "learning_rate": 2e-05, + "loss": 0.2162, + "step": 1728 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2198, + "step": 1729 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2585, + "step": 1730 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2326, + "step": 1731 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2569, + "step": 1732 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2063, + "step": 1733 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2544, + "step": 1734 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2249, + "step": 1735 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2301, + "step": 1736 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.1892, + "step": 1737 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.1763, + "step": 1738 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2629, + "step": 1739 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2136, + "step": 1740 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2137, + "step": 1741 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.235, + "step": 1742 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2296, + "step": 1743 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.1976, + "step": 1744 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2341, + "step": 1745 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.1843, + "step": 1746 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2167, + "step": 1747 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2074, + "step": 1748 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2077, + "step": 1749 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2134, + "step": 1750 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2556, + "step": 1751 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2223, + "step": 1752 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.1793, + "step": 1753 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2401, + "step": 1754 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2045, + "step": 1755 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2084, + "step": 1756 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.1939, + "step": 1757 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2493, + "step": 1758 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2258, + "step": 1759 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.1888, + "step": 1760 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2267, + "step": 1761 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2203, + "step": 1762 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2441, + "step": 1763 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2007, + "step": 1764 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2333, + "step": 1765 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2472, + "step": 1766 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2271, + "step": 1767 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 1768 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2099, + "step": 1769 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.1926, + "step": 1770 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2085, + "step": 1771 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2024, + "step": 1772 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2, + "step": 1773 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2085, + "step": 1774 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.1909, + "step": 1775 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2901, + "step": 1776 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.189, + "step": 1777 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2309, + "step": 1778 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2442, + "step": 1779 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 1780 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2025, + "step": 1781 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.189, + "step": 1782 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.1902, + "step": 1783 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2099, + "step": 1784 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2805, + "step": 1785 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.1891, + "step": 1786 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2772, + "step": 1787 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2684, + "step": 1788 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2244, + "step": 1789 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2212, + "step": 1790 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2722, + "step": 1791 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2292, + "step": 1792 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2472, + "step": 1793 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2264, + "step": 1794 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2155, + "step": 1795 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2347, + "step": 1796 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2167, + "step": 1797 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2131, + "step": 1798 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.1688, + "step": 1799 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2373, + "step": 1800 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2477, + "step": 1801 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.3073, + "step": 1802 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.192, + "step": 1803 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.1928, + "step": 1804 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.1837, + "step": 1805 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2442, + "step": 1806 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2343, + "step": 1807 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2177, + "step": 1808 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2168, + "step": 1809 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.1914, + "step": 1810 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.1942, + "step": 1811 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2361, + "step": 1812 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2314, + "step": 1813 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.1756, + "step": 1814 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.1983, + "step": 1815 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.1993, + "step": 1816 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.1696, + "step": 1817 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.1947, + "step": 1818 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2433, + "step": 1819 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2547, + "step": 1820 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2419, + "step": 1821 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2592, + "step": 1822 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2052, + "step": 1823 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.1626, + "step": 1824 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 1825 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.1605, + "step": 1826 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2068, + "step": 1827 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2355, + "step": 1828 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2771, + "step": 1829 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.1803, + "step": 1830 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2287, + "step": 1831 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.173, + "step": 1832 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.1966, + "step": 1833 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2318, + "step": 1834 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 1835 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2301, + "step": 1836 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2374, + "step": 1837 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2635, + "step": 1838 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2348, + "step": 1839 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.191, + "step": 1840 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.1853, + "step": 1841 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.1758, + "step": 1842 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.1922, + "step": 1843 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.1853, + "step": 1844 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2304, + "step": 1845 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.1999, + "step": 1846 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.211, + "step": 1847 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2413, + "step": 1848 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2292, + "step": 1849 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.1864, + "step": 1850 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.183, + "step": 1851 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2413, + "step": 1852 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.236, + "step": 1853 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2223, + "step": 1854 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2879, + "step": 1855 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2276, + "step": 1856 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.1988, + "step": 1857 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2104, + "step": 1858 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.243, + "step": 1859 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2267, + "step": 1860 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.1918, + "step": 1861 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2017, + "step": 1862 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2279, + "step": 1863 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2481, + "step": 1864 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2681, + "step": 1865 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 1866 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2746, + "step": 1867 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.1705, + "step": 1868 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2545, + "step": 1869 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2187, + "step": 1870 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2227, + "step": 1871 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.1932, + "step": 1872 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.1998, + "step": 1873 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.1811, + "step": 1874 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2258, + "step": 1875 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2364, + "step": 1876 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.185, + "step": 1877 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2197, + "step": 1878 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2021, + "step": 1879 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2185, + "step": 1880 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2282, + "step": 1881 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.239, + "step": 1882 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.184, + "step": 1883 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2676, + "step": 1884 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2249, + "step": 1885 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.1583, + "step": 1886 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.26, + "step": 1887 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.1955, + "step": 1888 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2747, + "step": 1889 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2396, + "step": 1890 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2053, + "step": 1891 + }, + { + "epoch": 0.11, + "learning_rate": 2e-05, + "loss": 0.2365, + "step": 1892 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2339, + "step": 1893 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.3041, + "step": 1894 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1791, + "step": 1895 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2411, + "step": 1896 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1973, + "step": 1897 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2387, + "step": 1898 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1909, + "step": 1899 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2195, + "step": 1900 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2102, + "step": 1901 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2708, + "step": 1902 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2341, + "step": 1903 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2175, + "step": 1904 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2312, + "step": 1905 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2299, + "step": 1906 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1896, + "step": 1907 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2062, + "step": 1908 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2325, + "step": 1909 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2049, + "step": 1910 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1983, + "step": 1911 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2253, + "step": 1912 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1899, + "step": 1913 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2148, + "step": 1914 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1978, + "step": 1915 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1874, + "step": 1916 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.199, + "step": 1917 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2322, + "step": 1918 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1848, + "step": 1919 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2165, + "step": 1920 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1735, + "step": 1921 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2014, + "step": 1922 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2145, + "step": 1923 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2206, + "step": 1924 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 1925 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2139, + "step": 1926 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2339, + "step": 1927 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1873, + "step": 1928 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1763, + "step": 1929 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2519, + "step": 1930 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2059, + "step": 1931 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1699, + "step": 1932 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2504, + "step": 1933 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2098, + "step": 1934 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1925, + "step": 1935 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2086, + "step": 1936 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1782, + "step": 1937 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2573, + "step": 1938 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2202, + "step": 1939 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2111, + "step": 1940 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2354, + "step": 1941 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2426, + "step": 1942 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1819, + "step": 1943 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 1944 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.256, + "step": 1945 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2172, + "step": 1946 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2519, + "step": 1947 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2491, + "step": 1948 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2454, + "step": 1949 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.228, + "step": 1950 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1936, + "step": 1951 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.193, + "step": 1952 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.23, + "step": 1953 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1913, + "step": 1954 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2037, + "step": 1955 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.235, + "step": 1956 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2642, + "step": 1957 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1978, + "step": 1958 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1901, + "step": 1959 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2181, + "step": 1960 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1675, + "step": 1961 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1713, + "step": 1962 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1987, + "step": 1963 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.222, + "step": 1964 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2959, + "step": 1965 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2447, + "step": 1966 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2045, + "step": 1967 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2152, + "step": 1968 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2457, + "step": 1969 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2338, + "step": 1970 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2151, + "step": 1971 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2186, + "step": 1972 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2635, + "step": 1973 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2338, + "step": 1974 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2319, + "step": 1975 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2158, + "step": 1976 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2732, + "step": 1977 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2193, + "step": 1978 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2098, + "step": 1979 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.216, + "step": 1980 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1858, + "step": 1981 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2036, + "step": 1982 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.173, + "step": 1983 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2082, + "step": 1984 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2273, + "step": 1985 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2141, + "step": 1986 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2273, + "step": 1987 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1657, + "step": 1988 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2389, + "step": 1989 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2515, + "step": 1990 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2947, + "step": 1991 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2255, + "step": 1992 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.19, + "step": 1993 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1995, + "step": 1994 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2125, + "step": 1995 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2161, + "step": 1996 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2342, + "step": 1997 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2413, + "step": 1998 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2423, + "step": 1999 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1849, + "step": 2000 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2167, + "step": 2001 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1948, + "step": 2002 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2399, + "step": 2003 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2194, + "step": 2004 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2068, + "step": 2005 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2348, + "step": 2006 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2478, + "step": 2007 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2162, + "step": 2008 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2116, + "step": 2009 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1883, + "step": 2010 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 2011 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2301, + "step": 2012 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2201, + "step": 2013 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2435, + "step": 2014 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2128, + "step": 2015 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1602, + "step": 2016 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2154, + "step": 2017 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1862, + "step": 2018 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1719, + "step": 2019 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2119, + "step": 2020 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2238, + "step": 2021 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.201, + "step": 2022 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2081, + "step": 2023 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2231, + "step": 2024 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1817, + "step": 2025 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2385, + "step": 2026 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1904, + "step": 2027 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2628, + "step": 2028 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1945, + "step": 2029 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1866, + "step": 2030 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1969, + "step": 2031 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1755, + "step": 2032 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2397, + "step": 2033 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2299, + "step": 2034 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2631, + "step": 2035 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2177, + "step": 2036 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1978, + "step": 2037 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2026, + "step": 2038 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2052, + "step": 2039 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1703, + "step": 2040 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2312, + "step": 2041 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.201, + "step": 2042 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1848, + "step": 2043 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1757, + "step": 2044 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1788, + "step": 2045 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2113, + "step": 2046 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1497, + "step": 2047 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2319, + "step": 2048 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2143, + "step": 2049 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.242, + "step": 2050 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2003, + "step": 2051 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1771, + "step": 2052 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2153, + "step": 2053 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1857, + "step": 2054 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.1986, + "step": 2055 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2608, + "step": 2056 + }, + { + "epoch": 0.12, + "learning_rate": 2e-05, + "loss": 0.2414, + "step": 2057 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2392, + "step": 2058 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 2059 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1905, + "step": 2060 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1896, + "step": 2061 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2484, + "step": 2062 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2574, + "step": 2063 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 2064 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2384, + "step": 2065 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1855, + "step": 2066 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 2067 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.201, + "step": 2068 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1729, + "step": 2069 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1892, + "step": 2070 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2777, + "step": 2071 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1958, + "step": 2072 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2003, + "step": 2073 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2131, + "step": 2074 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2359, + "step": 2075 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1722, + "step": 2076 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1914, + "step": 2077 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2109, + "step": 2078 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2243, + "step": 2079 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1864, + "step": 2080 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.256, + "step": 2081 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 2082 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1577, + "step": 2083 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2116, + "step": 2084 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2356, + "step": 2085 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2332, + "step": 2086 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2286, + "step": 2087 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1842, + "step": 2088 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1975, + "step": 2089 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2046, + "step": 2090 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2496, + "step": 2091 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1968, + "step": 2092 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2132, + "step": 2093 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.191, + "step": 2094 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2245, + "step": 2095 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1803, + "step": 2096 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 2097 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2035, + "step": 2098 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1696, + "step": 2099 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2134, + "step": 2100 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2003, + "step": 2101 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.246, + "step": 2102 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1837, + "step": 2103 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.179, + "step": 2104 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1928, + "step": 2105 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1852, + "step": 2106 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2445, + "step": 2107 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1723, + "step": 2108 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1702, + "step": 2109 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1883, + "step": 2110 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2211, + "step": 2111 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2249, + "step": 2112 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2684, + "step": 2113 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 2114 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1908, + "step": 2115 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2415, + "step": 2116 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1769, + "step": 2117 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2103, + "step": 2118 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2096, + "step": 2119 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 2120 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.19, + "step": 2121 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1697, + "step": 2122 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2461, + "step": 2123 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.273, + "step": 2124 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2303, + "step": 2125 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1868, + "step": 2126 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2136, + "step": 2127 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1945, + "step": 2128 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2394, + "step": 2129 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1988, + "step": 2130 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2382, + "step": 2131 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2051, + "step": 2132 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 2133 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.222, + "step": 2134 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.203, + "step": 2135 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1798, + "step": 2136 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2463, + "step": 2137 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1954, + "step": 2138 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.231, + "step": 2139 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1895, + "step": 2140 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2073, + "step": 2141 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1963, + "step": 2142 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1698, + "step": 2143 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1928, + "step": 2144 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1713, + "step": 2145 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2615, + "step": 2146 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1944, + "step": 2147 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2503, + "step": 2148 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1915, + "step": 2149 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.244, + "step": 2150 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2168, + "step": 2151 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2642, + "step": 2152 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 2153 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1961, + "step": 2154 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1839, + "step": 2155 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2209, + "step": 2156 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2018, + "step": 2157 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2139, + "step": 2158 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2195, + "step": 2159 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1834, + "step": 2160 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1839, + "step": 2161 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2412, + "step": 2162 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2077, + "step": 2163 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1913, + "step": 2164 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1927, + "step": 2165 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1645, + "step": 2166 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2258, + "step": 2167 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2166, + "step": 2168 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2313, + "step": 2169 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2229, + "step": 2170 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2022, + "step": 2171 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2101, + "step": 2172 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1661, + "step": 2173 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2704, + "step": 2174 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1805, + "step": 2175 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2936, + "step": 2176 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1732, + "step": 2177 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2485, + "step": 2178 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2066, + "step": 2179 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1888, + "step": 2180 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2147, + "step": 2181 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1861, + "step": 2182 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2681, + "step": 2183 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1902, + "step": 2184 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2217, + "step": 2185 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1816, + "step": 2186 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.255, + "step": 2187 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1945, + "step": 2188 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2281, + "step": 2189 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1994, + "step": 2190 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1988, + "step": 2191 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2011, + "step": 2192 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2147, + "step": 2193 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1787, + "step": 2194 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1868, + "step": 2195 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2172, + "step": 2196 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2367, + "step": 2197 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1722, + "step": 2198 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1817, + "step": 2199 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2099, + "step": 2200 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.217, + "step": 2201 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1912, + "step": 2202 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.205, + "step": 2203 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2146, + "step": 2204 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2264, + "step": 2205 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1707, + "step": 2206 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1813, + "step": 2207 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1954, + "step": 2208 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1942, + "step": 2209 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 2210 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1857, + "step": 2211 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2292, + "step": 2212 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2336, + "step": 2213 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2268, + "step": 2214 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2181, + "step": 2215 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1906, + "step": 2216 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.1899, + "step": 2217 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.236, + "step": 2218 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.251, + "step": 2219 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2188, + "step": 2220 + }, + { + "epoch": 0.13, + "learning_rate": 2e-05, + "loss": 0.2198, + "step": 2221 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1632, + "step": 2222 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2098, + "step": 2223 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 2224 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2198, + "step": 2225 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1695, + "step": 2226 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1963, + "step": 2227 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.201, + "step": 2228 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1926, + "step": 2229 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2228, + "step": 2230 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2402, + "step": 2231 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1922, + "step": 2232 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1975, + "step": 2233 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1911, + "step": 2234 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2375, + "step": 2235 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1687, + "step": 2236 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1799, + "step": 2237 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1846, + "step": 2238 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2023, + "step": 2239 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1979, + "step": 2240 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2067, + "step": 2241 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1924, + "step": 2242 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2228, + "step": 2243 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2158, + "step": 2244 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2145, + "step": 2245 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2091, + "step": 2246 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2112, + "step": 2247 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2214, + "step": 2248 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.148, + "step": 2249 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2143, + "step": 2250 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.183, + "step": 2251 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2088, + "step": 2252 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2261, + "step": 2253 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1935, + "step": 2254 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1972, + "step": 2255 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1807, + "step": 2256 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1595, + "step": 2257 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2179, + "step": 2258 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1877, + "step": 2259 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1666, + "step": 2260 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.188, + "step": 2261 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2242, + "step": 2262 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.182, + "step": 2263 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2305, + "step": 2264 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2268, + "step": 2265 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2512, + "step": 2266 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2175, + "step": 2267 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1978, + "step": 2268 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2702, + "step": 2269 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1944, + "step": 2270 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1879, + "step": 2271 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2174, + "step": 2272 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1935, + "step": 2273 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.205, + "step": 2274 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2358, + "step": 2275 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1952, + "step": 2276 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 2277 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2307, + "step": 2278 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1792, + "step": 2279 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2162, + "step": 2280 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2058, + "step": 2281 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1684, + "step": 2282 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2045, + "step": 2283 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2544, + "step": 2284 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2393, + "step": 2285 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2112, + "step": 2286 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2392, + "step": 2287 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2433, + "step": 2288 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2132, + "step": 2289 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2002, + "step": 2290 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1835, + "step": 2291 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2665, + "step": 2292 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2152, + "step": 2293 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 2294 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.224, + "step": 2295 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2259, + "step": 2296 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1697, + "step": 2297 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2185, + "step": 2298 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1953, + "step": 2299 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2425, + "step": 2300 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1797, + "step": 2301 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2212, + "step": 2302 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2525, + "step": 2303 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1968, + "step": 2304 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2308, + "step": 2305 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 2306 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1637, + "step": 2307 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1922, + "step": 2308 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1994, + "step": 2309 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1915, + "step": 2310 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1882, + "step": 2311 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2462, + "step": 2312 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2395, + "step": 2313 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2486, + "step": 2314 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.201, + "step": 2315 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1937, + "step": 2316 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.191, + "step": 2317 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2015, + "step": 2318 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2462, + "step": 2319 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1807, + "step": 2320 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1894, + "step": 2321 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1607, + "step": 2322 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.22, + "step": 2323 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2026, + "step": 2324 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2216, + "step": 2325 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2416, + "step": 2326 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2291, + "step": 2327 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1913, + "step": 2328 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1983, + "step": 2329 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1755, + "step": 2330 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2843, + "step": 2331 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2144, + "step": 2332 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1902, + "step": 2333 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2181, + "step": 2334 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2528, + "step": 2335 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2807, + "step": 2336 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 2337 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2121, + "step": 2338 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1756, + "step": 2339 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1907, + "step": 2340 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1979, + "step": 2341 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1872, + "step": 2342 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2213, + "step": 2343 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2604, + "step": 2344 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1803, + "step": 2345 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2392, + "step": 2346 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1834, + "step": 2347 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1921, + "step": 2348 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1898, + "step": 2349 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2156, + "step": 2350 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 2351 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2288, + "step": 2352 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2172, + "step": 2353 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1756, + "step": 2354 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2485, + "step": 2355 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1996, + "step": 2356 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1793, + "step": 2357 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2152, + "step": 2358 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1761, + "step": 2359 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1626, + "step": 2360 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 2361 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2136, + "step": 2362 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2066, + "step": 2363 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2032, + "step": 2364 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1891, + "step": 2365 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1634, + "step": 2366 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2117, + "step": 2367 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1669, + "step": 2368 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1801, + "step": 2369 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2107, + "step": 2370 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1622, + "step": 2371 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1936, + "step": 2372 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2137, + "step": 2373 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1758, + "step": 2374 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2032, + "step": 2375 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2021, + "step": 2376 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1639, + "step": 2377 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2349, + "step": 2378 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1849, + "step": 2379 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2157, + "step": 2380 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1955, + "step": 2381 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.1634, + "step": 2382 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2141, + "step": 2383 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2121, + "step": 2384 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2225, + "step": 2385 + }, + { + "epoch": 0.14, + "learning_rate": 2e-05, + "loss": 0.2245, + "step": 2386 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2075, + "step": 2387 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2485, + "step": 2388 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 2389 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1848, + "step": 2390 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1884, + "step": 2391 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1855, + "step": 2392 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.267, + "step": 2393 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2085, + "step": 2394 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1848, + "step": 2395 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1786, + "step": 2396 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1781, + "step": 2397 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2371, + "step": 2398 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2364, + "step": 2399 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2329, + "step": 2400 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1953, + "step": 2401 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2177, + "step": 2402 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1646, + "step": 2403 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.174, + "step": 2404 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2097, + "step": 2405 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2152, + "step": 2406 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1653, + "step": 2407 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2119, + "step": 2408 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2066, + "step": 2409 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2129, + "step": 2410 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2097, + "step": 2411 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1634, + "step": 2412 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2282, + "step": 2413 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2436, + "step": 2414 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1743, + "step": 2415 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2131, + "step": 2416 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2148, + "step": 2417 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1689, + "step": 2418 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2361, + "step": 2419 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.209, + "step": 2420 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2171, + "step": 2421 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2383, + "step": 2422 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2452, + "step": 2423 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2107, + "step": 2424 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2867, + "step": 2425 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2218, + "step": 2426 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2123, + "step": 2427 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.201, + "step": 2428 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1756, + "step": 2429 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1988, + "step": 2430 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2535, + "step": 2431 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2076, + "step": 2432 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2019, + "step": 2433 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2081, + "step": 2434 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1851, + "step": 2435 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1943, + "step": 2436 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2174, + "step": 2437 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1841, + "step": 2438 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.199, + "step": 2439 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2485, + "step": 2440 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2172, + "step": 2441 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.253, + "step": 2442 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2423, + "step": 2443 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2174, + "step": 2444 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2047, + "step": 2445 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2207, + "step": 2446 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2053, + "step": 2447 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1975, + "step": 2448 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1977, + "step": 2449 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2638, + "step": 2450 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1845, + "step": 2451 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 2452 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2132, + "step": 2453 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2009, + "step": 2454 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.193, + "step": 2455 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1769, + "step": 2456 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1736, + "step": 2457 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2274, + "step": 2458 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 2459 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.195, + "step": 2460 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1849, + "step": 2461 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1816, + "step": 2462 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2055, + "step": 2463 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2038, + "step": 2464 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1789, + "step": 2465 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2278, + "step": 2466 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2102, + "step": 2467 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2277, + "step": 2468 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2675, + "step": 2469 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2174, + "step": 2470 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1989, + "step": 2471 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2099, + "step": 2472 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1874, + "step": 2473 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2022, + "step": 2474 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1658, + "step": 2475 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1781, + "step": 2476 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2433, + "step": 2477 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2434, + "step": 2478 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2182, + "step": 2479 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1622, + "step": 2480 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2129, + "step": 2481 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2247, + "step": 2482 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1944, + "step": 2483 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1505, + "step": 2484 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2112, + "step": 2485 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1786, + "step": 2486 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1814, + "step": 2487 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1962, + "step": 2488 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2246, + "step": 2489 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1823, + "step": 2490 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.236, + "step": 2491 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2044, + "step": 2492 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2124, + "step": 2493 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 2494 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1812, + "step": 2495 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2067, + "step": 2496 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2157, + "step": 2497 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2138, + "step": 2498 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2312, + "step": 2499 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2098, + "step": 2500 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1876, + "step": 2501 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 2502 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1772, + "step": 2503 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 2504 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2352, + "step": 2505 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2326, + "step": 2506 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1837, + "step": 2507 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2666, + "step": 2508 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1853, + "step": 2509 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1521, + "step": 2510 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1815, + "step": 2511 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2411, + "step": 2512 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2024, + "step": 2513 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.206, + "step": 2514 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2254, + "step": 2515 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1695, + "step": 2516 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2572, + "step": 2517 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2208, + "step": 2518 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 2519 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2429, + "step": 2520 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2602, + "step": 2521 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2089, + "step": 2522 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2667, + "step": 2523 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1706, + "step": 2524 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.278, + "step": 2525 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1762, + "step": 2526 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2204, + "step": 2527 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1928, + "step": 2528 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2185, + "step": 2529 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2429, + "step": 2530 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1914, + "step": 2531 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1869, + "step": 2532 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2411, + "step": 2533 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2056, + "step": 2534 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1727, + "step": 2535 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2332, + "step": 2536 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1979, + "step": 2537 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 2538 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1681, + "step": 2539 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1794, + "step": 2540 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2613, + "step": 2541 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2168, + "step": 2542 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1757, + "step": 2543 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.185, + "step": 2544 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.231, + "step": 2545 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1904, + "step": 2546 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.195, + "step": 2547 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2254, + "step": 2548 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.2129, + "step": 2549 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.223, + "step": 2550 + }, + { + "epoch": 0.15, + "learning_rate": 2e-05, + "loss": 0.1931, + "step": 2551 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.189, + "step": 2552 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1951, + "step": 2553 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 2554 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1913, + "step": 2555 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1919, + "step": 2556 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2385, + "step": 2557 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2076, + "step": 2558 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2084, + "step": 2559 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2146, + "step": 2560 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2106, + "step": 2561 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2259, + "step": 2562 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1805, + "step": 2563 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.199, + "step": 2564 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2313, + "step": 2565 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1798, + "step": 2566 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1937, + "step": 2567 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2059, + "step": 2568 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2071, + "step": 2569 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2237, + "step": 2570 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2001, + "step": 2571 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1942, + "step": 2572 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2192, + "step": 2573 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1722, + "step": 2574 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1805, + "step": 2575 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2001, + "step": 2576 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2327, + "step": 2577 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1948, + "step": 2578 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1726, + "step": 2579 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2472, + "step": 2580 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2164, + "step": 2581 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2162, + "step": 2582 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2322, + "step": 2583 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1703, + "step": 2584 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1822, + "step": 2585 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2221, + "step": 2586 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2346, + "step": 2587 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.199, + "step": 2588 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1885, + "step": 2589 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1823, + "step": 2590 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1949, + "step": 2591 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2284, + "step": 2592 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2291, + "step": 2593 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2959, + "step": 2594 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2182, + "step": 2595 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2424, + "step": 2596 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1876, + "step": 2597 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2108, + "step": 2598 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1938, + "step": 2599 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2123, + "step": 2600 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1963, + "step": 2601 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2135, + "step": 2602 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1876, + "step": 2603 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 2604 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1883, + "step": 2605 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1573, + "step": 2606 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2233, + "step": 2607 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2304, + "step": 2608 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1742, + "step": 2609 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2552, + "step": 2610 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1758, + "step": 2611 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2073, + "step": 2612 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2147, + "step": 2613 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1834, + "step": 2614 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2119, + "step": 2615 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2607, + "step": 2616 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2317, + "step": 2617 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2072, + "step": 2618 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1787, + "step": 2619 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1928, + "step": 2620 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.236, + "step": 2621 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2165, + "step": 2622 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2011, + "step": 2623 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2054, + "step": 2624 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1638, + "step": 2625 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 2626 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2368, + "step": 2627 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1998, + "step": 2628 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2402, + "step": 2629 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1979, + "step": 2630 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1846, + "step": 2631 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2481, + "step": 2632 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2303, + "step": 2633 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1908, + "step": 2634 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2189, + "step": 2635 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1829, + "step": 2636 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2385, + "step": 2637 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2061, + "step": 2638 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1999, + "step": 2639 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2068, + "step": 2640 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2277, + "step": 2641 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2156, + "step": 2642 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1809, + "step": 2643 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2467, + "step": 2644 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1859, + "step": 2645 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2323, + "step": 2646 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1579, + "step": 2647 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1705, + "step": 2648 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1622, + "step": 2649 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2133, + "step": 2650 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2331, + "step": 2651 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.165, + "step": 2652 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2227, + "step": 2653 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2656, + "step": 2654 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1949, + "step": 2655 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1773, + "step": 2656 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2529, + "step": 2657 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2534, + "step": 2658 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2227, + "step": 2659 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1746, + "step": 2660 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1685, + "step": 2661 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2404, + "step": 2662 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2024, + "step": 2663 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2396, + "step": 2664 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1986, + "step": 2665 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2738, + "step": 2666 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1987, + "step": 2667 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2644, + "step": 2668 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1806, + "step": 2669 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2688, + "step": 2670 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2022, + "step": 2671 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 2672 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1655, + "step": 2673 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2103, + "step": 2674 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.202, + "step": 2675 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1926, + "step": 2676 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2142, + "step": 2677 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2313, + "step": 2678 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2271, + "step": 2679 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2311, + "step": 2680 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.249, + "step": 2681 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1872, + "step": 2682 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2168, + "step": 2683 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.239, + "step": 2684 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1814, + "step": 2685 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2337, + "step": 2686 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 2687 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1855, + "step": 2688 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2303, + "step": 2689 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2207, + "step": 2690 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2399, + "step": 2691 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2373, + "step": 2692 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2185, + "step": 2693 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2735, + "step": 2694 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1797, + "step": 2695 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2311, + "step": 2696 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1993, + "step": 2697 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2506, + "step": 2698 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2156, + "step": 2699 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2157, + "step": 2700 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1502, + "step": 2701 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1816, + "step": 2702 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2085, + "step": 2703 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.196, + "step": 2704 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2364, + "step": 2705 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2132, + "step": 2706 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2043, + "step": 2707 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.243, + "step": 2708 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2303, + "step": 2709 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1756, + "step": 2710 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2223, + "step": 2711 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1906, + "step": 2712 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2196, + "step": 2713 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.1845, + "step": 2714 + }, + { + "epoch": 0.16, + "learning_rate": 2e-05, + "loss": 0.2497, + "step": 2715 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1758, + "step": 2716 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2353, + "step": 2717 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1939, + "step": 2718 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1883, + "step": 2719 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 2720 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1977, + "step": 2721 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1597, + "step": 2722 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2397, + "step": 2723 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.239, + "step": 2724 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1738, + "step": 2725 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2346, + "step": 2726 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2156, + "step": 2727 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1824, + "step": 2728 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2235, + "step": 2729 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2529, + "step": 2730 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2377, + "step": 2731 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2032, + "step": 2732 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2114, + "step": 2733 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2355, + "step": 2734 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1993, + "step": 2735 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1731, + "step": 2736 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2494, + "step": 2737 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2485, + "step": 2738 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1616, + "step": 2739 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1871, + "step": 2740 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2206, + "step": 2741 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1815, + "step": 2742 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2563, + "step": 2743 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1846, + "step": 2744 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.183, + "step": 2745 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1754, + "step": 2746 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1968, + "step": 2747 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2018, + "step": 2748 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 2749 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2082, + "step": 2750 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1765, + "step": 2751 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1789, + "step": 2752 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1538, + "step": 2753 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2545, + "step": 2754 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2375, + "step": 2755 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1773, + "step": 2756 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1935, + "step": 2757 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1711, + "step": 2758 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2513, + "step": 2759 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2101, + "step": 2760 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.212, + "step": 2761 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1863, + "step": 2762 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 2763 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1691, + "step": 2764 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2046, + "step": 2765 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2242, + "step": 2766 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1824, + "step": 2767 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2106, + "step": 2768 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1914, + "step": 2769 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2015, + "step": 2770 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1898, + "step": 2771 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2189, + "step": 2772 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1797, + "step": 2773 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1693, + "step": 2774 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1867, + "step": 2775 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2698, + "step": 2776 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1983, + "step": 2777 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.3052, + "step": 2778 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1927, + "step": 2779 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.242, + "step": 2780 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2236, + "step": 2781 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2165, + "step": 2782 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1704, + "step": 2783 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1688, + "step": 2784 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2019, + "step": 2785 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2214, + "step": 2786 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1948, + "step": 2787 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2151, + "step": 2788 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1816, + "step": 2789 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2337, + "step": 2790 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.207, + "step": 2791 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1931, + "step": 2792 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1693, + "step": 2793 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2315, + "step": 2794 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2146, + "step": 2795 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2134, + "step": 2796 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2335, + "step": 2797 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2349, + "step": 2798 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1741, + "step": 2799 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1851, + "step": 2800 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2345, + "step": 2801 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2304, + "step": 2802 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1869, + "step": 2803 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 2804 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1925, + "step": 2805 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2499, + "step": 2806 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2159, + "step": 2807 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1856, + "step": 2808 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1935, + "step": 2809 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1973, + "step": 2810 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.208, + "step": 2811 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2061, + "step": 2812 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.234, + "step": 2813 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2216, + "step": 2814 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1799, + "step": 2815 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.238, + "step": 2816 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1896, + "step": 2817 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2145, + "step": 2818 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1973, + "step": 2819 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2165, + "step": 2820 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1928, + "step": 2821 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.212, + "step": 2822 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2296, + "step": 2823 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1801, + "step": 2824 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2139, + "step": 2825 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2251, + "step": 2826 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1794, + "step": 2827 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1911, + "step": 2828 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2092, + "step": 2829 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2263, + "step": 2830 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2035, + "step": 2831 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2025, + "step": 2832 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1861, + "step": 2833 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1821, + "step": 2834 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1941, + "step": 2835 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2239, + "step": 2836 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1779, + "step": 2837 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2192, + "step": 2838 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2479, + "step": 2839 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1966, + "step": 2840 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1893, + "step": 2841 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2552, + "step": 2842 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.191, + "step": 2843 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2419, + "step": 2844 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2138, + "step": 2845 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1552, + "step": 2846 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1694, + "step": 2847 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2014, + "step": 2848 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2055, + "step": 2849 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.211, + "step": 2850 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1604, + "step": 2851 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2085, + "step": 2852 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2349, + "step": 2853 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1804, + "step": 2854 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2146, + "step": 2855 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2012, + "step": 2856 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2226, + "step": 2857 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2513, + "step": 2858 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 2859 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2104, + "step": 2860 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 2861 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2063, + "step": 2862 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1837, + "step": 2863 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1852, + "step": 2864 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.205, + "step": 2865 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.182, + "step": 2866 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2707, + "step": 2867 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1804, + "step": 2868 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1689, + "step": 2869 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1747, + "step": 2870 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2134, + "step": 2871 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2209, + "step": 2872 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2374, + "step": 2873 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1847, + "step": 2874 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2249, + "step": 2875 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2174, + "step": 2876 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.2089, + "step": 2877 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1705, + "step": 2878 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 2879 + }, + { + "epoch": 0.17, + "learning_rate": 2e-05, + "loss": 0.1588, + "step": 2880 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1956, + "step": 2881 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2149, + "step": 2882 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2332, + "step": 2883 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2187, + "step": 2884 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2073, + "step": 2885 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1848, + "step": 2886 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.189, + "step": 2887 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.171, + "step": 2888 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2019, + "step": 2889 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2461, + "step": 2890 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2098, + "step": 2891 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2027, + "step": 2892 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2211, + "step": 2893 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1992, + "step": 2894 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1779, + "step": 2895 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2244, + "step": 2896 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2425, + "step": 2897 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1795, + "step": 2898 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2236, + "step": 2899 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1679, + "step": 2900 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1908, + "step": 2901 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1753, + "step": 2902 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1967, + "step": 2903 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1954, + "step": 2904 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1925, + "step": 2905 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1976, + "step": 2906 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2056, + "step": 2907 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2117, + "step": 2908 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1855, + "step": 2909 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2341, + "step": 2910 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1973, + "step": 2911 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1792, + "step": 2912 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2181, + "step": 2913 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1673, + "step": 2914 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2148, + "step": 2915 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 2916 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1921, + "step": 2917 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.201, + "step": 2918 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1827, + "step": 2919 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2489, + "step": 2920 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.226, + "step": 2921 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2117, + "step": 2922 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.203, + "step": 2923 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1992, + "step": 2924 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2116, + "step": 2925 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1631, + "step": 2926 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2384, + "step": 2927 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2814, + "step": 2928 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1953, + "step": 2929 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1988, + "step": 2930 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2564, + "step": 2931 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2577, + "step": 2932 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2298, + "step": 2933 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2494, + "step": 2934 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.173, + "step": 2935 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2301, + "step": 2936 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2386, + "step": 2937 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1713, + "step": 2938 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2059, + "step": 2939 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1985, + "step": 2940 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2324, + "step": 2941 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.222, + "step": 2942 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1798, + "step": 2943 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1947, + "step": 2944 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2557, + "step": 2945 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1965, + "step": 2946 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2523, + "step": 2947 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1899, + "step": 2948 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2401, + "step": 2949 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.242, + "step": 2950 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1795, + "step": 2951 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2538, + "step": 2952 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.154, + "step": 2953 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2201, + "step": 2954 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1792, + "step": 2955 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2003, + "step": 2956 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2059, + "step": 2957 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.293, + "step": 2958 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1987, + "step": 2959 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2176, + "step": 2960 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2299, + "step": 2961 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.229, + "step": 2962 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2287, + "step": 2963 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.19, + "step": 2964 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2162, + "step": 2965 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1929, + "step": 2966 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2121, + "step": 2967 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1703, + "step": 2968 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2112, + "step": 2969 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 2970 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1819, + "step": 2971 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1712, + "step": 2972 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2122, + "step": 2973 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1773, + "step": 2974 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2258, + "step": 2975 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2117, + "step": 2976 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 2977 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2239, + "step": 2978 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.192, + "step": 2979 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1824, + "step": 2980 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.158, + "step": 2981 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 2982 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2019, + "step": 2983 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1777, + "step": 2984 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1926, + "step": 2985 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1851, + "step": 2986 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2121, + "step": 2987 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2183, + "step": 2988 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1767, + "step": 2989 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.187, + "step": 2990 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2078, + "step": 2991 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2067, + "step": 2992 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1597, + "step": 2993 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1784, + "step": 2994 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1952, + "step": 2995 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2679, + "step": 2996 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2472, + "step": 2997 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1927, + "step": 2998 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1554, + "step": 2999 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2655, + "step": 3000 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2248, + "step": 3001 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2148, + "step": 3002 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1889, + "step": 3003 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1553, + "step": 3004 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1696, + "step": 3005 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2053, + "step": 3006 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2199, + "step": 3007 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1806, + "step": 3008 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1938, + "step": 3009 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2115, + "step": 3010 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1595, + "step": 3011 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.255, + "step": 3012 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1914, + "step": 3013 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 3014 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2106, + "step": 3015 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1682, + "step": 3016 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2526, + "step": 3017 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1733, + "step": 3018 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2141, + "step": 3019 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2169, + "step": 3020 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1993, + "step": 3021 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2216, + "step": 3022 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1758, + "step": 3023 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1763, + "step": 3024 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1961, + "step": 3025 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1845, + "step": 3026 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1588, + "step": 3027 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1912, + "step": 3028 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.177, + "step": 3029 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2206, + "step": 3030 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.216, + "step": 3031 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1879, + "step": 3032 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1725, + "step": 3033 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2036, + "step": 3034 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2382, + "step": 3035 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2074, + "step": 3036 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1819, + "step": 3037 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2095, + "step": 3038 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2453, + "step": 3039 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1776, + "step": 3040 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.2255, + "step": 3041 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.239, + "step": 3042 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1884, + "step": 3043 + }, + { + "epoch": 0.18, + "learning_rate": 2e-05, + "loss": 0.1752, + "step": 3044 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2218, + "step": 3045 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1726, + "step": 3046 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1855, + "step": 3047 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2034, + "step": 3048 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2068, + "step": 3049 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 3050 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2239, + "step": 3051 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1832, + "step": 3052 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1785, + "step": 3053 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2356, + "step": 3054 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2069, + "step": 3055 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2151, + "step": 3056 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1488, + "step": 3057 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2516, + "step": 3058 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1631, + "step": 3059 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1766, + "step": 3060 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2057, + "step": 3061 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2171, + "step": 3062 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1968, + "step": 3063 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2201, + "step": 3064 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1805, + "step": 3065 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.223, + "step": 3066 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2014, + "step": 3067 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2267, + "step": 3068 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.182, + "step": 3069 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 3070 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1689, + "step": 3071 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1722, + "step": 3072 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2471, + "step": 3073 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.252, + "step": 3074 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2314, + "step": 3075 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1734, + "step": 3076 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2033, + "step": 3077 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.218, + "step": 3078 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1984, + "step": 3079 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2127, + "step": 3080 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2573, + "step": 3081 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2226, + "step": 3082 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1836, + "step": 3083 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.164, + "step": 3084 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2387, + "step": 3085 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2275, + "step": 3086 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2723, + "step": 3087 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2018, + "step": 3088 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1809, + "step": 3089 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1616, + "step": 3090 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1886, + "step": 3091 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2184, + "step": 3092 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.202, + "step": 3093 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2283, + "step": 3094 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2001, + "step": 3095 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2508, + "step": 3096 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2141, + "step": 3097 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1772, + "step": 3098 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1994, + "step": 3099 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1942, + "step": 3100 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1632, + "step": 3101 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2199, + "step": 3102 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.191, + "step": 3103 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1958, + "step": 3104 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.174, + "step": 3105 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1978, + "step": 3106 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1841, + "step": 3107 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2124, + "step": 3108 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1772, + "step": 3109 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.24, + "step": 3110 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2014, + "step": 3111 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2114, + "step": 3112 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2148, + "step": 3113 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 3114 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2118, + "step": 3115 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2464, + "step": 3116 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2201, + "step": 3117 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1842, + "step": 3118 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1911, + "step": 3119 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2169, + "step": 3120 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2088, + "step": 3121 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1919, + "step": 3122 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1796, + "step": 3123 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1722, + "step": 3124 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2118, + "step": 3125 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2071, + "step": 3126 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2144, + "step": 3127 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1558, + "step": 3128 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2021, + "step": 3129 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2372, + "step": 3130 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1808, + "step": 3131 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2189, + "step": 3132 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2536, + "step": 3133 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2748, + "step": 3134 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2683, + "step": 3135 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2025, + "step": 3136 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2506, + "step": 3137 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1683, + "step": 3138 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1578, + "step": 3139 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2431, + "step": 3140 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1763, + "step": 3141 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1899, + "step": 3142 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2335, + "step": 3143 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1934, + "step": 3144 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1834, + "step": 3145 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2426, + "step": 3146 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1673, + "step": 3147 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2069, + "step": 3148 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1702, + "step": 3149 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2295, + "step": 3150 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1786, + "step": 3151 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1828, + "step": 3152 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.183, + "step": 3153 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1918, + "step": 3154 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1962, + "step": 3155 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 3156 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1852, + "step": 3157 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2126, + "step": 3158 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2222, + "step": 3159 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.167, + "step": 3160 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 3161 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2054, + "step": 3162 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 3163 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2028, + "step": 3164 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1728, + "step": 3165 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2053, + "step": 3166 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1879, + "step": 3167 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2065, + "step": 3168 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2381, + "step": 3169 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1656, + "step": 3170 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1822, + "step": 3171 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1861, + "step": 3172 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1485, + "step": 3173 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1806, + "step": 3174 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1966, + "step": 3175 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2492, + "step": 3176 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2451, + "step": 3177 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1966, + "step": 3178 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1911, + "step": 3179 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.177, + "step": 3180 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1877, + "step": 3181 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2195, + "step": 3182 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1852, + "step": 3183 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.252, + "step": 3184 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 3185 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2127, + "step": 3186 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1755, + "step": 3187 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.168, + "step": 3188 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2083, + "step": 3189 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1853, + "step": 3190 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2318, + "step": 3191 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2208, + "step": 3192 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2339, + "step": 3193 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2355, + "step": 3194 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2246, + "step": 3195 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1825, + "step": 3196 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2391, + "step": 3197 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1588, + "step": 3198 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1757, + "step": 3199 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2217, + "step": 3200 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2238, + "step": 3201 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1968, + "step": 3202 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1753, + "step": 3203 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1714, + "step": 3204 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1666, + "step": 3205 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.2421, + "step": 3206 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1643, + "step": 3207 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1962, + "step": 3208 + }, + { + "epoch": 0.19, + "learning_rate": 2e-05, + "loss": 0.1835, + "step": 3209 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2022, + "step": 3210 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1531, + "step": 3211 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2003, + "step": 3212 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2511, + "step": 3213 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1724, + "step": 3214 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1836, + "step": 3215 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 3216 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1885, + "step": 3217 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1634, + "step": 3218 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1979, + "step": 3219 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2173, + "step": 3220 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2138, + "step": 3221 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2372, + "step": 3222 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 3223 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2453, + "step": 3224 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1928, + "step": 3225 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2198, + "step": 3226 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1629, + "step": 3227 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2074, + "step": 3228 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2074, + "step": 3229 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1831, + "step": 3230 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2119, + "step": 3231 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1922, + "step": 3232 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1605, + "step": 3233 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2445, + "step": 3234 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2283, + "step": 3235 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1795, + "step": 3236 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1632, + "step": 3237 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2182, + "step": 3238 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2277, + "step": 3239 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2192, + "step": 3240 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1725, + "step": 3241 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2306, + "step": 3242 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2526, + "step": 3243 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2372, + "step": 3244 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2161, + "step": 3245 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1985, + "step": 3246 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2381, + "step": 3247 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1677, + "step": 3248 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1896, + "step": 3249 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2067, + "step": 3250 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1788, + "step": 3251 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1796, + "step": 3252 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1649, + "step": 3253 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1641, + "step": 3254 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2415, + "step": 3255 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1874, + "step": 3256 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1743, + "step": 3257 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2195, + "step": 3258 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1749, + "step": 3259 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2074, + "step": 3260 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 3261 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2948, + "step": 3262 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2515, + "step": 3263 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2075, + "step": 3264 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1805, + "step": 3265 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2056, + "step": 3266 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1833, + "step": 3267 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2555, + "step": 3268 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1915, + "step": 3269 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1726, + "step": 3270 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.184, + "step": 3271 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1658, + "step": 3272 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2171, + "step": 3273 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2089, + "step": 3274 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1642, + "step": 3275 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2517, + "step": 3276 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.3072, + "step": 3277 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1816, + "step": 3278 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2096, + "step": 3279 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2251, + "step": 3280 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1841, + "step": 3281 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1987, + "step": 3282 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 3283 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2084, + "step": 3284 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1941, + "step": 3285 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2353, + "step": 3286 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1967, + "step": 3287 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1926, + "step": 3288 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1733, + "step": 3289 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2574, + "step": 3290 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 3291 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2056, + "step": 3292 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1889, + "step": 3293 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.24, + "step": 3294 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2349, + "step": 3295 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2382, + "step": 3296 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2226, + "step": 3297 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2273, + "step": 3298 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2016, + "step": 3299 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2233, + "step": 3300 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.193, + "step": 3301 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2022, + "step": 3302 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2069, + "step": 3303 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2094, + "step": 3304 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2105, + "step": 3305 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1752, + "step": 3306 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1772, + "step": 3307 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2066, + "step": 3308 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2194, + "step": 3309 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.169, + "step": 3310 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1807, + "step": 3311 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1787, + "step": 3312 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2001, + "step": 3313 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1981, + "step": 3314 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1793, + "step": 3315 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1757, + "step": 3316 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2284, + "step": 3317 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2553, + "step": 3318 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1946, + "step": 3319 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.214, + "step": 3320 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1962, + "step": 3321 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2189, + "step": 3322 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2341, + "step": 3323 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1979, + "step": 3324 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1578, + "step": 3325 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2045, + "step": 3326 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 3327 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1989, + "step": 3328 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2246, + "step": 3329 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2169, + "step": 3330 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2312, + "step": 3331 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2172, + "step": 3332 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2444, + "step": 3333 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2204, + "step": 3334 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2765, + "step": 3335 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2467, + "step": 3336 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.235, + "step": 3337 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1848, + "step": 3338 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2034, + "step": 3339 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1986, + "step": 3340 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2687, + "step": 3341 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1963, + "step": 3342 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2014, + "step": 3343 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1663, + "step": 3344 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2094, + "step": 3345 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1909, + "step": 3346 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1579, + "step": 3347 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2172, + "step": 3348 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1948, + "step": 3349 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.164, + "step": 3350 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.202, + "step": 3351 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.3068, + "step": 3352 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1875, + "step": 3353 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1944, + "step": 3354 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2074, + "step": 3355 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2096, + "step": 3356 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2632, + "step": 3357 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1862, + "step": 3358 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 3359 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2655, + "step": 3360 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1689, + "step": 3361 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1934, + "step": 3362 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1923, + "step": 3363 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2015, + "step": 3364 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2164, + "step": 3365 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2133, + "step": 3366 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1973, + "step": 3367 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1922, + "step": 3368 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1837, + "step": 3369 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.1972, + "step": 3370 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.194, + "step": 3371 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2285, + "step": 3372 + }, + { + "epoch": 0.2, + "learning_rate": 2e-05, + "loss": 0.2244, + "step": 3373 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2579, + "step": 3374 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1972, + "step": 3375 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1827, + "step": 3376 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2072, + "step": 3377 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2176, + "step": 3378 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1755, + "step": 3379 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2131, + "step": 3380 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1908, + "step": 3381 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1834, + "step": 3382 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2036, + "step": 3383 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1802, + "step": 3384 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2505, + "step": 3385 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1935, + "step": 3386 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.176, + "step": 3387 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1817, + "step": 3388 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2514, + "step": 3389 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 3390 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2055, + "step": 3391 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1812, + "step": 3392 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.172, + "step": 3393 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2323, + "step": 3394 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2119, + "step": 3395 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2304, + "step": 3396 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.169, + "step": 3397 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.248, + "step": 3398 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2206, + "step": 3399 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2202, + "step": 3400 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2233, + "step": 3401 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2091, + "step": 3402 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1746, + "step": 3403 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1728, + "step": 3404 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 3405 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 3406 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2155, + "step": 3407 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1729, + "step": 3408 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2424, + "step": 3409 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2283, + "step": 3410 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2016, + "step": 3411 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1925, + "step": 3412 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.152, + "step": 3413 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1973, + "step": 3414 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.182, + "step": 3415 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2155, + "step": 3416 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.232, + "step": 3417 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1804, + "step": 3418 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2233, + "step": 3419 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2093, + "step": 3420 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2084, + "step": 3421 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 3422 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2019, + "step": 3423 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1682, + "step": 3424 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1773, + "step": 3425 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.23, + "step": 3426 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2199, + "step": 3427 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1851, + "step": 3428 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2053, + "step": 3429 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1914, + "step": 3430 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1965, + "step": 3431 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1524, + "step": 3432 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2195, + "step": 3433 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1675, + "step": 3434 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1999, + "step": 3435 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1746, + "step": 3436 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.238, + "step": 3437 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2376, + "step": 3438 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2249, + "step": 3439 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1888, + "step": 3440 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1883, + "step": 3441 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2035, + "step": 3442 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2093, + "step": 3443 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1869, + "step": 3444 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2303, + "step": 3445 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 3446 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1662, + "step": 3447 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2358, + "step": 3448 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1701, + "step": 3449 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1854, + "step": 3450 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2674, + "step": 3451 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 3452 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2136, + "step": 3453 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2089, + "step": 3454 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1802, + "step": 3455 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1634, + "step": 3456 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2434, + "step": 3457 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1754, + "step": 3458 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2027, + "step": 3459 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1736, + "step": 3460 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2658, + "step": 3461 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1987, + "step": 3462 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2259, + "step": 3463 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2131, + "step": 3464 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2206, + "step": 3465 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1881, + "step": 3466 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1858, + "step": 3467 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2087, + "step": 3468 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1683, + "step": 3469 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1795, + "step": 3470 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1655, + "step": 3471 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1995, + "step": 3472 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1732, + "step": 3473 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.225, + "step": 3474 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.192, + "step": 3475 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 3476 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1676, + "step": 3477 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.159, + "step": 3478 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 3479 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2279, + "step": 3480 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1913, + "step": 3481 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1923, + "step": 3482 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1873, + "step": 3483 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2056, + "step": 3484 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1932, + "step": 3485 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1879, + "step": 3486 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2177, + "step": 3487 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1728, + "step": 3488 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2063, + "step": 3489 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2014, + "step": 3490 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2086, + "step": 3491 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1712, + "step": 3492 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1706, + "step": 3493 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.208, + "step": 3494 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2015, + "step": 3495 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.235, + "step": 3496 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1879, + "step": 3497 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1788, + "step": 3498 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2127, + "step": 3499 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1668, + "step": 3500 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 3501 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2125, + "step": 3502 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1868, + "step": 3503 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1606, + "step": 3504 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2537, + "step": 3505 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2197, + "step": 3506 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2417, + "step": 3507 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2665, + "step": 3508 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1582, + "step": 3509 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1579, + "step": 3510 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2329, + "step": 3511 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2066, + "step": 3512 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1796, + "step": 3513 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1947, + "step": 3514 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2402, + "step": 3515 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1857, + "step": 3516 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1834, + "step": 3517 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.196, + "step": 3518 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1985, + "step": 3519 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1909, + "step": 3520 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1782, + "step": 3521 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 3522 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1877, + "step": 3523 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2155, + "step": 3524 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1771, + "step": 3525 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.202, + "step": 3526 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2151, + "step": 3527 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1784, + "step": 3528 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.201, + "step": 3529 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1651, + "step": 3530 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1498, + "step": 3531 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2151, + "step": 3532 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.187, + "step": 3533 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1743, + "step": 3534 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1787, + "step": 3535 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.1937, + "step": 3536 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.2231, + "step": 3537 + }, + { + "epoch": 0.21, + "learning_rate": 2e-05, + "loss": 0.241, + "step": 3538 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1793, + "step": 3539 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2405, + "step": 3540 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1735, + "step": 3541 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1443, + "step": 3542 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2344, + "step": 3543 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2224, + "step": 3544 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2188, + "step": 3545 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2242, + "step": 3546 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.152, + "step": 3547 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2363, + "step": 3548 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2163, + "step": 3549 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1688, + "step": 3550 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.218, + "step": 3551 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1858, + "step": 3552 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 3553 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2057, + "step": 3554 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1588, + "step": 3555 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1932, + "step": 3556 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1831, + "step": 3557 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2162, + "step": 3558 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1669, + "step": 3559 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.174, + "step": 3560 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1693, + "step": 3561 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1743, + "step": 3562 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2623, + "step": 3563 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1921, + "step": 3564 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2115, + "step": 3565 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2459, + "step": 3566 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1821, + "step": 3567 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2019, + "step": 3568 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2206, + "step": 3569 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1885, + "step": 3570 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1734, + "step": 3571 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1706, + "step": 3572 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1735, + "step": 3573 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2087, + "step": 3574 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1652, + "step": 3575 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2336, + "step": 3576 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1734, + "step": 3577 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1893, + "step": 3578 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1561, + "step": 3579 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2102, + "step": 3580 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1677, + "step": 3581 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2043, + "step": 3582 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2222, + "step": 3583 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2031, + "step": 3584 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1843, + "step": 3585 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1855, + "step": 3586 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2052, + "step": 3587 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2042, + "step": 3588 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1787, + "step": 3589 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2354, + "step": 3590 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1876, + "step": 3591 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1694, + "step": 3592 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1831, + "step": 3593 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1997, + "step": 3594 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2511, + "step": 3595 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1928, + "step": 3596 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2199, + "step": 3597 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 3598 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1743, + "step": 3599 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2192, + "step": 3600 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2175, + "step": 3601 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1945, + "step": 3602 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1732, + "step": 3603 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1897, + "step": 3604 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1958, + "step": 3605 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2331, + "step": 3606 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2605, + "step": 3607 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1854, + "step": 3608 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1568, + "step": 3609 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.268, + "step": 3610 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1795, + "step": 3611 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1507, + "step": 3612 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.218, + "step": 3613 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1758, + "step": 3614 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1682, + "step": 3615 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1904, + "step": 3616 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2047, + "step": 3617 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2312, + "step": 3618 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2133, + "step": 3619 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1678, + "step": 3620 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.148, + "step": 3621 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2164, + "step": 3622 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1776, + "step": 3623 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.235, + "step": 3624 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2831, + "step": 3625 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1921, + "step": 3626 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 3627 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2301, + "step": 3628 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2018, + "step": 3629 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1776, + "step": 3630 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.189, + "step": 3631 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2198, + "step": 3632 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2007, + "step": 3633 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2543, + "step": 3634 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2264, + "step": 3635 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1651, + "step": 3636 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1963, + "step": 3637 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2404, + "step": 3638 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2233, + "step": 3639 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1894, + "step": 3640 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2221, + "step": 3641 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1931, + "step": 3642 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.168, + "step": 3643 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1866, + "step": 3644 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1546, + "step": 3645 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1884, + "step": 3646 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2213, + "step": 3647 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2131, + "step": 3648 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1924, + "step": 3649 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1517, + "step": 3650 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1984, + "step": 3651 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2027, + "step": 3652 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1757, + "step": 3653 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2091, + "step": 3654 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1775, + "step": 3655 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1814, + "step": 3656 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1783, + "step": 3657 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1969, + "step": 3658 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1907, + "step": 3659 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2131, + "step": 3660 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1741, + "step": 3661 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1825, + "step": 3662 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1525, + "step": 3663 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2213, + "step": 3664 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1956, + "step": 3665 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1946, + "step": 3666 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2273, + "step": 3667 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1769, + "step": 3668 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.19, + "step": 3669 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1772, + "step": 3670 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1918, + "step": 3671 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2634, + "step": 3672 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2452, + "step": 3673 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2277, + "step": 3674 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1749, + "step": 3675 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1593, + "step": 3676 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2381, + "step": 3677 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2335, + "step": 3678 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2479, + "step": 3679 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2044, + "step": 3680 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1684, + "step": 3681 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1688, + "step": 3682 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1741, + "step": 3683 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2212, + "step": 3684 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 3685 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1729, + "step": 3686 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1905, + "step": 3687 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2053, + "step": 3688 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2163, + "step": 3689 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.159, + "step": 3690 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1729, + "step": 3691 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1988, + "step": 3692 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2015, + "step": 3693 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1602, + "step": 3694 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2131, + "step": 3695 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2045, + "step": 3696 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2028, + "step": 3697 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1799, + "step": 3698 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1777, + "step": 3699 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2325, + "step": 3700 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.1759, + "step": 3701 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.179, + "step": 3702 + }, + { + "epoch": 0.22, + "learning_rate": 2e-05, + "loss": 0.2066, + "step": 3703 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2451, + "step": 3704 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.204, + "step": 3705 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2131, + "step": 3706 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2733, + "step": 3707 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2378, + "step": 3708 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1495, + "step": 3709 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2381, + "step": 3710 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1906, + "step": 3711 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1794, + "step": 3712 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1888, + "step": 3713 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2066, + "step": 3714 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1751, + "step": 3715 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1777, + "step": 3716 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2154, + "step": 3717 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2291, + "step": 3718 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1975, + "step": 3719 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1781, + "step": 3720 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1853, + "step": 3721 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1733, + "step": 3722 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1807, + "step": 3723 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.225, + "step": 3724 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.231, + "step": 3725 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2013, + "step": 3726 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.155, + "step": 3727 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1699, + "step": 3728 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 3729 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2462, + "step": 3730 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2369, + "step": 3731 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2133, + "step": 3732 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1952, + "step": 3733 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1588, + "step": 3734 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1853, + "step": 3735 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2069, + "step": 3736 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2399, + "step": 3737 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1611, + "step": 3738 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1846, + "step": 3739 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1953, + "step": 3740 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1773, + "step": 3741 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2138, + "step": 3742 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1944, + "step": 3743 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2294, + "step": 3744 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1532, + "step": 3745 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2281, + "step": 3746 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2738, + "step": 3747 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1969, + "step": 3748 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1773, + "step": 3749 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2068, + "step": 3750 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1753, + "step": 3751 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2572, + "step": 3752 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1805, + "step": 3753 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1737, + "step": 3754 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2687, + "step": 3755 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1721, + "step": 3756 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2339, + "step": 3757 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1808, + "step": 3758 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1688, + "step": 3759 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2038, + "step": 3760 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1739, + "step": 3761 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2109, + "step": 3762 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2124, + "step": 3763 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2476, + "step": 3764 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2056, + "step": 3765 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1798, + "step": 3766 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1685, + "step": 3767 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 3768 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.178, + "step": 3769 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2227, + "step": 3770 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1863, + "step": 3771 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1904, + "step": 3772 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1688, + "step": 3773 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 3774 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1847, + "step": 3775 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1847, + "step": 3776 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2169, + "step": 3777 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2011, + "step": 3778 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2598, + "step": 3779 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1854, + "step": 3780 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 3781 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2057, + "step": 3782 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 3783 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2144, + "step": 3784 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.193, + "step": 3785 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.204, + "step": 3786 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1705, + "step": 3787 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2338, + "step": 3788 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2027, + "step": 3789 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2185, + "step": 3790 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1518, + "step": 3791 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1468, + "step": 3792 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2191, + "step": 3793 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1827, + "step": 3794 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.175, + "step": 3795 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2083, + "step": 3796 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1906, + "step": 3797 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2276, + "step": 3798 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1801, + "step": 3799 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2377, + "step": 3800 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2151, + "step": 3801 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2111, + "step": 3802 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2213, + "step": 3803 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2067, + "step": 3804 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1532, + "step": 3805 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1701, + "step": 3806 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1903, + "step": 3807 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1808, + "step": 3808 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1869, + "step": 3809 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1987, + "step": 3810 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2013, + "step": 3811 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2275, + "step": 3812 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1687, + "step": 3813 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2419, + "step": 3814 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 3815 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1903, + "step": 3816 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1852, + "step": 3817 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2224, + "step": 3818 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1747, + "step": 3819 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2096, + "step": 3820 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2006, + "step": 3821 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.259, + "step": 3822 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2355, + "step": 3823 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2023, + "step": 3824 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2263, + "step": 3825 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1623, + "step": 3826 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1877, + "step": 3827 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1622, + "step": 3828 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2176, + "step": 3829 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2329, + "step": 3830 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.203, + "step": 3831 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1902, + "step": 3832 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.189, + "step": 3833 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1678, + "step": 3834 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1847, + "step": 3835 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1928, + "step": 3836 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2279, + "step": 3837 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1782, + "step": 3838 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1984, + "step": 3839 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2478, + "step": 3840 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1621, + "step": 3841 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 3842 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 3843 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1881, + "step": 3844 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 3845 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.206, + "step": 3846 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2078, + "step": 3847 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2118, + "step": 3848 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2151, + "step": 3849 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1823, + "step": 3850 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.244, + "step": 3851 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2412, + "step": 3852 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1845, + "step": 3853 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2128, + "step": 3854 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2099, + "step": 3855 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1956, + "step": 3856 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1589, + "step": 3857 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2239, + "step": 3858 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2088, + "step": 3859 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1843, + "step": 3860 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2193, + "step": 3861 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1711, + "step": 3862 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2209, + "step": 3863 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1814, + "step": 3864 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2213, + "step": 3865 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.2065, + "step": 3866 + }, + { + "epoch": 0.23, + "learning_rate": 2e-05, + "loss": 0.1664, + "step": 3867 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1826, + "step": 3868 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1778, + "step": 3869 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1787, + "step": 3870 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.225, + "step": 3871 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2033, + "step": 3872 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2184, + "step": 3873 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1963, + "step": 3874 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2571, + "step": 3875 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2311, + "step": 3876 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2127, + "step": 3877 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1894, + "step": 3878 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1687, + "step": 3879 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1797, + "step": 3880 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2232, + "step": 3881 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 3882 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2174, + "step": 3883 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1863, + "step": 3884 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2451, + "step": 3885 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1786, + "step": 3886 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 3887 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1965, + "step": 3888 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 3889 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 3890 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 3891 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1661, + "step": 3892 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2061, + "step": 3893 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1917, + "step": 3894 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1893, + "step": 3895 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2305, + "step": 3896 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1767, + "step": 3897 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2482, + "step": 3898 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 3899 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.205, + "step": 3900 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2115, + "step": 3901 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2008, + "step": 3902 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1734, + "step": 3903 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2106, + "step": 3904 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 3905 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1873, + "step": 3906 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1661, + "step": 3907 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2127, + "step": 3908 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2057, + "step": 3909 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1858, + "step": 3910 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1726, + "step": 3911 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1576, + "step": 3912 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2039, + "step": 3913 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 3914 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 3915 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.171, + "step": 3916 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.224, + "step": 3917 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2123, + "step": 3918 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2057, + "step": 3919 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1924, + "step": 3920 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1954, + "step": 3921 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2164, + "step": 3922 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1774, + "step": 3923 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1533, + "step": 3924 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.186, + "step": 3925 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2193, + "step": 3926 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1891, + "step": 3927 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 3928 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1762, + "step": 3929 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1942, + "step": 3930 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.179, + "step": 3931 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1742, + "step": 3932 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2374, + "step": 3933 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2062, + "step": 3934 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1775, + "step": 3935 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2074, + "step": 3936 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1697, + "step": 3937 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.212, + "step": 3938 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.208, + "step": 3939 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 3940 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1956, + "step": 3941 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1825, + "step": 3942 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.188, + "step": 3943 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1982, + "step": 3944 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 3945 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1841, + "step": 3946 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1967, + "step": 3947 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1699, + "step": 3948 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1595, + "step": 3949 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1933, + "step": 3950 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2066, + "step": 3951 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 3952 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2392, + "step": 3953 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1781, + "step": 3954 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2261, + "step": 3955 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.17, + "step": 3956 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.199, + "step": 3957 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1708, + "step": 3958 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2248, + "step": 3959 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1719, + "step": 3960 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.205, + "step": 3961 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2036, + "step": 3962 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2155, + "step": 3963 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1844, + "step": 3964 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1907, + "step": 3965 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1566, + "step": 3966 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1778, + "step": 3967 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2928, + "step": 3968 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2135, + "step": 3969 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1815, + "step": 3970 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2369, + "step": 3971 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2384, + "step": 3972 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2215, + "step": 3973 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1671, + "step": 3974 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1847, + "step": 3975 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.168, + "step": 3976 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1908, + "step": 3977 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1894, + "step": 3978 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1663, + "step": 3979 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2388, + "step": 3980 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1835, + "step": 3981 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2224, + "step": 3982 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2782, + "step": 3983 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1799, + "step": 3984 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1777, + "step": 3985 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.172, + "step": 3986 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2028, + "step": 3987 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1755, + "step": 3988 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2851, + "step": 3989 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1661, + "step": 3990 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1616, + "step": 3991 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.206, + "step": 3992 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1867, + "step": 3993 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1925, + "step": 3994 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1872, + "step": 3995 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1557, + "step": 3996 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.186, + "step": 3997 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1811, + "step": 3998 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1652, + "step": 3999 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2163, + "step": 4000 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2025, + "step": 4001 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.189, + "step": 4002 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1707, + "step": 4003 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2078, + "step": 4004 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1876, + "step": 4005 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1922, + "step": 4006 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1688, + "step": 4007 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1743, + "step": 4008 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1624, + "step": 4009 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2047, + "step": 4010 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1952, + "step": 4011 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2269, + "step": 4012 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1664, + "step": 4013 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1921, + "step": 4014 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1924, + "step": 4015 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.18, + "step": 4016 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 4017 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2114, + "step": 4018 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 4019 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1499, + "step": 4020 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2209, + "step": 4021 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2211, + "step": 4022 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1797, + "step": 4023 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 4024 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.19, + "step": 4025 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2197, + "step": 4026 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1829, + "step": 4027 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.2048, + "step": 4028 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.219, + "step": 4029 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1878, + "step": 4030 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1856, + "step": 4031 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.1811, + "step": 4032 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1917, + "step": 4033 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1969, + "step": 4034 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1785, + "step": 4035 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2002, + "step": 4036 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1774, + "step": 4037 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2132, + "step": 4038 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2253, + "step": 4039 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1864, + "step": 4040 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1682, + "step": 4041 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2488, + "step": 4042 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1886, + "step": 4043 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1724, + "step": 4044 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2039, + "step": 4045 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1799, + "step": 4046 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1795, + "step": 4047 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2209, + "step": 4048 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.269, + "step": 4049 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 4050 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1929, + "step": 4051 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1961, + "step": 4052 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2119, + "step": 4053 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2378, + "step": 4054 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1604, + "step": 4055 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1948, + "step": 4056 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1949, + "step": 4057 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1669, + "step": 4058 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2206, + "step": 4059 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1786, + "step": 4060 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1706, + "step": 4061 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1748, + "step": 4062 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 4063 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2478, + "step": 4064 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1979, + "step": 4065 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1651, + "step": 4066 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2008, + "step": 4067 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1668, + "step": 4068 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1607, + "step": 4069 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1732, + "step": 4070 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1772, + "step": 4071 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2421, + "step": 4072 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1739, + "step": 4073 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1703, + "step": 4074 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2027, + "step": 4075 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1614, + "step": 4076 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1813, + "step": 4077 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1779, + "step": 4078 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 4079 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2085, + "step": 4080 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1632, + "step": 4081 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1667, + "step": 4082 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2021, + "step": 4083 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1996, + "step": 4084 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2512, + "step": 4085 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2271, + "step": 4086 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2039, + "step": 4087 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1899, + "step": 4088 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.169, + "step": 4089 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2247, + "step": 4090 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2195, + "step": 4091 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 4092 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.24, + "step": 4093 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2636, + "step": 4094 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1551, + "step": 4095 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2124, + "step": 4096 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2342, + "step": 4097 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1879, + "step": 4098 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1899, + "step": 4099 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1806, + "step": 4100 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1541, + "step": 4101 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2502, + "step": 4102 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1897, + "step": 4103 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2073, + "step": 4104 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2349, + "step": 4105 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1723, + "step": 4106 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2466, + "step": 4107 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2161, + "step": 4108 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1856, + "step": 4109 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2027, + "step": 4110 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1995, + "step": 4111 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1589, + "step": 4112 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1915, + "step": 4113 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2571, + "step": 4114 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2362, + "step": 4115 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2039, + "step": 4116 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2554, + "step": 4117 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2212, + "step": 4118 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1739, + "step": 4119 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2197, + "step": 4120 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1838, + "step": 4121 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1952, + "step": 4122 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.161, + "step": 4123 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1871, + "step": 4124 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1805, + "step": 4125 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 4126 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1997, + "step": 4127 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2569, + "step": 4128 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1788, + "step": 4129 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1958, + "step": 4130 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1634, + "step": 4131 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1848, + "step": 4132 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2158, + "step": 4133 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1901, + "step": 4134 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2283, + "step": 4135 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1974, + "step": 4136 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 4137 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1637, + "step": 4138 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1833, + "step": 4139 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.148, + "step": 4140 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2314, + "step": 4141 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1792, + "step": 4142 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2204, + "step": 4143 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 4144 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2221, + "step": 4145 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.201, + "step": 4146 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1683, + "step": 4147 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2219, + "step": 4148 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2388, + "step": 4149 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2474, + "step": 4150 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 4151 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2043, + "step": 4152 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2114, + "step": 4153 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 4154 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.182, + "step": 4155 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2759, + "step": 4156 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2094, + "step": 4157 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1793, + "step": 4158 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1678, + "step": 4159 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1683, + "step": 4160 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1753, + "step": 4161 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.163, + "step": 4162 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 4163 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1966, + "step": 4164 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2849, + "step": 4165 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2153, + "step": 4166 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1649, + "step": 4167 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1533, + "step": 4168 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1844, + "step": 4169 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2053, + "step": 4170 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2334, + "step": 4171 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 4172 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1956, + "step": 4173 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.202, + "step": 4174 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1644, + "step": 4175 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1789, + "step": 4176 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1876, + "step": 4177 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1602, + "step": 4178 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2297, + "step": 4179 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.162, + "step": 4180 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1881, + "step": 4181 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2481, + "step": 4182 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1913, + "step": 4183 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2217, + "step": 4184 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1857, + "step": 4185 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 4186 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 4187 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2099, + "step": 4188 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2353, + "step": 4189 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2036, + "step": 4190 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1872, + "step": 4191 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1732, + "step": 4192 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 4193 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1574, + "step": 4194 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.2226, + "step": 4195 + }, + { + "epoch": 0.25, + "learning_rate": 2e-05, + "loss": 0.1871, + "step": 4196 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 4197 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1935, + "step": 4198 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1682, + "step": 4199 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2243, + "step": 4200 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2383, + "step": 4201 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 4202 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1738, + "step": 4203 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1834, + "step": 4204 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.172, + "step": 4205 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 4206 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1963, + "step": 4207 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2236, + "step": 4208 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2173, + "step": 4209 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1647, + "step": 4210 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1989, + "step": 4211 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1985, + "step": 4212 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1782, + "step": 4213 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1888, + "step": 4214 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1947, + "step": 4215 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2117, + "step": 4216 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.162, + "step": 4217 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1582, + "step": 4218 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1774, + "step": 4219 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1961, + "step": 4220 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1637, + "step": 4221 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1774, + "step": 4222 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.171, + "step": 4223 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2665, + "step": 4224 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1751, + "step": 4225 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1511, + "step": 4226 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2142, + "step": 4227 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1541, + "step": 4228 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2012, + "step": 4229 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2436, + "step": 4230 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1813, + "step": 4231 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1935, + "step": 4232 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.189, + "step": 4233 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 4234 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.176, + "step": 4235 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 4236 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1785, + "step": 4237 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1897, + "step": 4238 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2178, + "step": 4239 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1747, + "step": 4240 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1688, + "step": 4241 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2599, + "step": 4242 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1982, + "step": 4243 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1787, + "step": 4244 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1778, + "step": 4245 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 4246 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1954, + "step": 4247 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2206, + "step": 4248 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1691, + "step": 4249 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1682, + "step": 4250 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2069, + "step": 4251 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2368, + "step": 4252 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2186, + "step": 4253 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2119, + "step": 4254 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1738, + "step": 4255 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1524, + "step": 4256 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1935, + "step": 4257 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2045, + "step": 4258 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1928, + "step": 4259 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1955, + "step": 4260 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1928, + "step": 4261 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1678, + "step": 4262 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 4263 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 4264 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2061, + "step": 4265 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2054, + "step": 4266 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1786, + "step": 4267 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2465, + "step": 4268 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2227, + "step": 4269 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1813, + "step": 4270 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2312, + "step": 4271 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2125, + "step": 4272 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 4273 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.206, + "step": 4274 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 4275 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2065, + "step": 4276 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1764, + "step": 4277 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1961, + "step": 4278 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2234, + "step": 4279 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.231, + "step": 4280 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.24, + "step": 4281 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2041, + "step": 4282 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2117, + "step": 4283 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1714, + "step": 4284 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1682, + "step": 4285 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1774, + "step": 4286 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1932, + "step": 4287 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.229, + "step": 4288 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2251, + "step": 4289 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2078, + "step": 4290 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2002, + "step": 4291 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1797, + "step": 4292 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1531, + "step": 4293 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1831, + "step": 4294 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1955, + "step": 4295 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2066, + "step": 4296 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2131, + "step": 4297 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2054, + "step": 4298 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1914, + "step": 4299 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1748, + "step": 4300 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 4301 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 4302 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2047, + "step": 4303 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1938, + "step": 4304 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1893, + "step": 4305 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2291, + "step": 4306 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.192, + "step": 4307 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1567, + "step": 4308 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.201, + "step": 4309 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1843, + "step": 4310 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.183, + "step": 4311 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1925, + "step": 4312 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1776, + "step": 4313 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 4314 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2324, + "step": 4315 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1928, + "step": 4316 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2047, + "step": 4317 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1573, + "step": 4318 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1908, + "step": 4319 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2246, + "step": 4320 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2248, + "step": 4321 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2201, + "step": 4322 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1821, + "step": 4323 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1685, + "step": 4324 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1831, + "step": 4325 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1972, + "step": 4326 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.206, + "step": 4327 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1745, + "step": 4328 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1656, + "step": 4329 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1734, + "step": 4330 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2021, + "step": 4331 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.186, + "step": 4332 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1727, + "step": 4333 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2015, + "step": 4334 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1988, + "step": 4335 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2083, + "step": 4336 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 4337 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1607, + "step": 4338 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2236, + "step": 4339 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2161, + "step": 4340 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1835, + "step": 4341 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1835, + "step": 4342 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2314, + "step": 4343 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2048, + "step": 4344 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2081, + "step": 4345 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2056, + "step": 4346 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1521, + "step": 4347 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2231, + "step": 4348 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2145, + "step": 4349 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.222, + "step": 4350 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1868, + "step": 4351 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.245, + "step": 4352 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1716, + "step": 4353 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.175, + "step": 4354 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1621, + "step": 4355 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1856, + "step": 4356 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1617, + "step": 4357 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2591, + "step": 4358 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.2006, + "step": 4359 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1872, + "step": 4360 + }, + { + "epoch": 0.26, + "learning_rate": 2e-05, + "loss": 0.1904, + "step": 4361 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1965, + "step": 4362 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.173, + "step": 4363 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1853, + "step": 4364 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1713, + "step": 4365 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2247, + "step": 4366 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1982, + "step": 4367 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1863, + "step": 4368 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 4369 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1612, + "step": 4370 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2309, + "step": 4371 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2128, + "step": 4372 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2132, + "step": 4373 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1685, + "step": 4374 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.218, + "step": 4375 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1693, + "step": 4376 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2052, + "step": 4377 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1639, + "step": 4378 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2093, + "step": 4379 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2038, + "step": 4380 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2706, + "step": 4381 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1976, + "step": 4382 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2544, + "step": 4383 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1752, + "step": 4384 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 4385 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1823, + "step": 4386 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2145, + "step": 4387 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2239, + "step": 4388 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1662, + "step": 4389 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2069, + "step": 4390 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1844, + "step": 4391 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1718, + "step": 4392 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.189, + "step": 4393 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.167, + "step": 4394 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1644, + "step": 4395 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.224, + "step": 4396 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2011, + "step": 4397 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1752, + "step": 4398 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1835, + "step": 4399 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2026, + "step": 4400 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 4401 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1989, + "step": 4402 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2154, + "step": 4403 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2165, + "step": 4404 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1719, + "step": 4405 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2082, + "step": 4406 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1647, + "step": 4407 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1516, + "step": 4408 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2196, + "step": 4409 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1651, + "step": 4410 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1689, + "step": 4411 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 4412 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1977, + "step": 4413 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1693, + "step": 4414 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1564, + "step": 4415 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1832, + "step": 4416 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.155, + "step": 4417 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 4418 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2114, + "step": 4419 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1891, + "step": 4420 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2168, + "step": 4421 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1749, + "step": 4422 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2123, + "step": 4423 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1682, + "step": 4424 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2066, + "step": 4425 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2111, + "step": 4426 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1697, + "step": 4427 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2244, + "step": 4428 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1808, + "step": 4429 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2006, + "step": 4430 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1963, + "step": 4431 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1789, + "step": 4432 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2513, + "step": 4433 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1877, + "step": 4434 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2526, + "step": 4435 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1596, + "step": 4436 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.193, + "step": 4437 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2062, + "step": 4438 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2262, + "step": 4439 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1993, + "step": 4440 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1851, + "step": 4441 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2133, + "step": 4442 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2004, + "step": 4443 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1861, + "step": 4444 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1639, + "step": 4445 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1732, + "step": 4446 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.215, + "step": 4447 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2049, + "step": 4448 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2222, + "step": 4449 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.201, + "step": 4450 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1966, + "step": 4451 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2112, + "step": 4452 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2511, + "step": 4453 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1962, + "step": 4454 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2338, + "step": 4455 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1765, + "step": 4456 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 4457 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.195, + "step": 4458 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1671, + "step": 4459 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1994, + "step": 4460 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.236, + "step": 4461 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1973, + "step": 4462 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1739, + "step": 4463 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1819, + "step": 4464 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1868, + "step": 4465 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2107, + "step": 4466 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1732, + "step": 4467 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1847, + "step": 4468 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1842, + "step": 4469 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1828, + "step": 4470 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 4471 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2133, + "step": 4472 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2317, + "step": 4473 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 4474 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2258, + "step": 4475 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2063, + "step": 4476 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1625, + "step": 4477 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1629, + "step": 4478 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1688, + "step": 4479 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1614, + "step": 4480 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2215, + "step": 4481 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1706, + "step": 4482 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2052, + "step": 4483 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2112, + "step": 4484 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1762, + "step": 4485 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1912, + "step": 4486 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1894, + "step": 4487 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1983, + "step": 4488 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1765, + "step": 4489 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 4490 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1651, + "step": 4491 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1983, + "step": 4492 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1993, + "step": 4493 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1662, + "step": 4494 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2292, + "step": 4495 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1851, + "step": 4496 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1725, + "step": 4497 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2075, + "step": 4498 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1919, + "step": 4499 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.192, + "step": 4500 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1528, + "step": 4501 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1989, + "step": 4502 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1734, + "step": 4503 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2285, + "step": 4504 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 4505 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2192, + "step": 4506 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 4507 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1732, + "step": 4508 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1592, + "step": 4509 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2185, + "step": 4510 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1528, + "step": 4511 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1774, + "step": 4512 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2394, + "step": 4513 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2324, + "step": 4514 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1546, + "step": 4515 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1513, + "step": 4516 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.2114, + "step": 4517 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 4518 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1497, + "step": 4519 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1817, + "step": 4520 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1684, + "step": 4521 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 4522 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 4523 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1878, + "step": 4524 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 4525 + }, + { + "epoch": 0.27, + "learning_rate": 2e-05, + "loss": 0.1943, + "step": 4526 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2385, + "step": 4527 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1824, + "step": 4528 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 4529 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1815, + "step": 4530 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.207, + "step": 4531 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1602, + "step": 4532 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1782, + "step": 4533 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1895, + "step": 4534 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2241, + "step": 4535 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.206, + "step": 4536 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1756, + "step": 4537 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1518, + "step": 4538 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 4539 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.195, + "step": 4540 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1755, + "step": 4541 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1889, + "step": 4542 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 4543 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1668, + "step": 4544 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1868, + "step": 4545 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1535, + "step": 4546 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1792, + "step": 4547 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 4548 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1942, + "step": 4549 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1616, + "step": 4550 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1836, + "step": 4551 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1888, + "step": 4552 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2149, + "step": 4553 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 4554 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1741, + "step": 4555 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1941, + "step": 4556 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2563, + "step": 4557 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 4558 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1601, + "step": 4559 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1607, + "step": 4560 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1872, + "step": 4561 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2199, + "step": 4562 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1803, + "step": 4563 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1716, + "step": 4564 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1997, + "step": 4565 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2159, + "step": 4566 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2714, + "step": 4567 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1528, + "step": 4568 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1901, + "step": 4569 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2013, + "step": 4570 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2256, + "step": 4571 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2133, + "step": 4572 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2715, + "step": 4573 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1853, + "step": 4574 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2103, + "step": 4575 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2125, + "step": 4576 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2046, + "step": 4577 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2036, + "step": 4578 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2035, + "step": 4579 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1797, + "step": 4580 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1704, + "step": 4581 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 4582 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1853, + "step": 4583 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1846, + "step": 4584 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1752, + "step": 4585 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1943, + "step": 4586 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2163, + "step": 4587 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2061, + "step": 4588 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2045, + "step": 4589 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1871, + "step": 4590 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2309, + "step": 4591 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1693, + "step": 4592 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2368, + "step": 4593 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1951, + "step": 4594 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1984, + "step": 4595 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1832, + "step": 4596 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2216, + "step": 4597 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 4598 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1593, + "step": 4599 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1782, + "step": 4600 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1988, + "step": 4601 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2092, + "step": 4602 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1726, + "step": 4603 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 4604 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1846, + "step": 4605 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2228, + "step": 4606 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 4607 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.17, + "step": 4608 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1854, + "step": 4609 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.244, + "step": 4610 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 4611 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 4612 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.174, + "step": 4613 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1664, + "step": 4614 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1608, + "step": 4615 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1795, + "step": 4616 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1755, + "step": 4617 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1839, + "step": 4618 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2072, + "step": 4619 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1662, + "step": 4620 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1718, + "step": 4621 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2053, + "step": 4622 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1875, + "step": 4623 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2097, + "step": 4624 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1892, + "step": 4625 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1959, + "step": 4626 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1858, + "step": 4627 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1781, + "step": 4628 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1911, + "step": 4629 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2059, + "step": 4630 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1865, + "step": 4631 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1567, + "step": 4632 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2091, + "step": 4633 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2314, + "step": 4634 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 4635 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1871, + "step": 4636 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2459, + "step": 4637 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1994, + "step": 4638 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2082, + "step": 4639 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2158, + "step": 4640 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1858, + "step": 4641 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1772, + "step": 4642 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1814, + "step": 4643 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2068, + "step": 4644 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2083, + "step": 4645 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1578, + "step": 4646 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2093, + "step": 4647 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 4648 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2446, + "step": 4649 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 4650 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 4651 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2593, + "step": 4652 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1715, + "step": 4653 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2423, + "step": 4654 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1832, + "step": 4655 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1963, + "step": 4656 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2293, + "step": 4657 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2284, + "step": 4658 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 4659 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1925, + "step": 4660 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1666, + "step": 4661 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1961, + "step": 4662 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2005, + "step": 4663 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1659, + "step": 4664 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1651, + "step": 4665 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.165, + "step": 4666 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1798, + "step": 4667 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1735, + "step": 4668 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2346, + "step": 4669 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.182, + "step": 4670 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2159, + "step": 4671 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.228, + "step": 4672 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1674, + "step": 4673 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1858, + "step": 4674 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1836, + "step": 4675 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1498, + "step": 4676 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2418, + "step": 4677 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1776, + "step": 4678 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2268, + "step": 4679 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1555, + "step": 4680 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2069, + "step": 4681 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2109, + "step": 4682 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.286, + "step": 4683 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1873, + "step": 4684 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2343, + "step": 4685 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1645, + "step": 4686 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1513, + "step": 4687 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.1654, + "step": 4688 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.2194, + "step": 4689 + }, + { + "epoch": 0.28, + "learning_rate": 2e-05, + "loss": 0.243, + "step": 4690 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1932, + "step": 4691 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1629, + "step": 4692 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1886, + "step": 4693 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1818, + "step": 4694 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1608, + "step": 4695 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 4696 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1759, + "step": 4697 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1912, + "step": 4698 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2117, + "step": 4699 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1658, + "step": 4700 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1699, + "step": 4701 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1954, + "step": 4702 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2098, + "step": 4703 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1727, + "step": 4704 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1549, + "step": 4705 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1901, + "step": 4706 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1761, + "step": 4707 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2031, + "step": 4708 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1849, + "step": 4709 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1994, + "step": 4710 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1924, + "step": 4711 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1941, + "step": 4712 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1606, + "step": 4713 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1884, + "step": 4714 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2182, + "step": 4715 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1668, + "step": 4716 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1821, + "step": 4717 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.206, + "step": 4718 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.191, + "step": 4719 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1602, + "step": 4720 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 4721 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 4722 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1549, + "step": 4723 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.198, + "step": 4724 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2107, + "step": 4725 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2365, + "step": 4726 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1789, + "step": 4727 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1949, + "step": 4728 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2031, + "step": 4729 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.182, + "step": 4730 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1861, + "step": 4731 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1757, + "step": 4732 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2134, + "step": 4733 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1912, + "step": 4734 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2229, + "step": 4735 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.156, + "step": 4736 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1905, + "step": 4737 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2005, + "step": 4738 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2255, + "step": 4739 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1766, + "step": 4740 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2251, + "step": 4741 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 4742 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1877, + "step": 4743 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2181, + "step": 4744 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 4745 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2065, + "step": 4746 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2006, + "step": 4747 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1839, + "step": 4748 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2017, + "step": 4749 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1997, + "step": 4750 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1842, + "step": 4751 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 4752 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2063, + "step": 4753 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2103, + "step": 4754 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2161, + "step": 4755 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2012, + "step": 4756 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2145, + "step": 4757 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2049, + "step": 4758 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2069, + "step": 4759 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1932, + "step": 4760 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1648, + "step": 4761 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2032, + "step": 4762 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1856, + "step": 4763 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1671, + "step": 4764 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1645, + "step": 4765 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2112, + "step": 4766 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1771, + "step": 4767 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1854, + "step": 4768 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1443, + "step": 4769 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1837, + "step": 4770 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2073, + "step": 4771 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 4772 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1762, + "step": 4773 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2265, + "step": 4774 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1765, + "step": 4775 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1672, + "step": 4776 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1582, + "step": 4777 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1855, + "step": 4778 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1814, + "step": 4779 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2178, + "step": 4780 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2358, + "step": 4781 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 4782 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.182, + "step": 4783 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1998, + "step": 4784 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2049, + "step": 4785 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 4786 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1762, + "step": 4787 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1855, + "step": 4788 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1777, + "step": 4789 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2215, + "step": 4790 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1798, + "step": 4791 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1686, + "step": 4792 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1726, + "step": 4793 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 4794 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 4795 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2203, + "step": 4796 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2137, + "step": 4797 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1884, + "step": 4798 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1857, + "step": 4799 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1792, + "step": 4800 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1518, + "step": 4801 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1668, + "step": 4802 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 4803 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 4804 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1932, + "step": 4805 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1867, + "step": 4806 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2051, + "step": 4807 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2004, + "step": 4808 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2058, + "step": 4809 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2057, + "step": 4810 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1727, + "step": 4811 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1585, + "step": 4812 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1847, + "step": 4813 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 4814 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2098, + "step": 4815 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1726, + "step": 4816 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1886, + "step": 4817 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1785, + "step": 4818 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2188, + "step": 4819 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1817, + "step": 4820 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1719, + "step": 4821 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2254, + "step": 4822 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.171, + "step": 4823 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1723, + "step": 4824 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1929, + "step": 4825 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 4826 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1984, + "step": 4827 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 4828 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1744, + "step": 4829 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1974, + "step": 4830 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1755, + "step": 4831 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1749, + "step": 4832 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2064, + "step": 4833 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1764, + "step": 4834 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2041, + "step": 4835 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1666, + "step": 4836 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1984, + "step": 4837 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.201, + "step": 4838 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1713, + "step": 4839 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 4840 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2227, + "step": 4841 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.16, + "step": 4842 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2265, + "step": 4843 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 4844 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1979, + "step": 4845 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2043, + "step": 4846 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1785, + "step": 4847 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1798, + "step": 4848 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1766, + "step": 4849 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.164, + "step": 4850 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1883, + "step": 4851 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1866, + "step": 4852 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1534, + "step": 4853 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.1915, + "step": 4854 + }, + { + "epoch": 0.29, + "learning_rate": 2e-05, + "loss": 0.2507, + "step": 4855 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 4856 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.165, + "step": 4857 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1867, + "step": 4858 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2283, + "step": 4859 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1635, + "step": 4860 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1943, + "step": 4861 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1947, + "step": 4862 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1945, + "step": 4863 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2042, + "step": 4864 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.184, + "step": 4865 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.18, + "step": 4866 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.21, + "step": 4867 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1616, + "step": 4868 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2083, + "step": 4869 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2052, + "step": 4870 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1771, + "step": 4871 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 4872 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1918, + "step": 4873 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1866, + "step": 4874 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2031, + "step": 4875 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2115, + "step": 4876 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 4877 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1926, + "step": 4878 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2048, + "step": 4879 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1593, + "step": 4880 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1704, + "step": 4881 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1744, + "step": 4882 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.203, + "step": 4883 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 4884 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2127, + "step": 4885 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1882, + "step": 4886 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 4887 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2237, + "step": 4888 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2083, + "step": 4889 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2352, + "step": 4890 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1834, + "step": 4891 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2018, + "step": 4892 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1841, + "step": 4893 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2305, + "step": 4894 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 4895 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2128, + "step": 4896 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1961, + "step": 4897 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2506, + "step": 4898 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2067, + "step": 4899 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1831, + "step": 4900 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1708, + "step": 4901 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2097, + "step": 4902 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 4903 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2132, + "step": 4904 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2661, + "step": 4905 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1702, + "step": 4906 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1807, + "step": 4907 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2142, + "step": 4908 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1967, + "step": 4909 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2141, + "step": 4910 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1841, + "step": 4911 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1598, + "step": 4912 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1647, + "step": 4913 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2181, + "step": 4914 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.218, + "step": 4915 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.208, + "step": 4916 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1888, + "step": 4917 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1742, + "step": 4918 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1685, + "step": 4919 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2019, + "step": 4920 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2179, + "step": 4921 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1926, + "step": 4922 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1649, + "step": 4923 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 4924 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1772, + "step": 4925 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2372, + "step": 4926 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1648, + "step": 4927 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1671, + "step": 4928 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 4929 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 4930 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1527, + "step": 4931 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 4932 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2158, + "step": 4933 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 4934 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.22, + "step": 4935 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2002, + "step": 4936 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1838, + "step": 4937 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 4938 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1914, + "step": 4939 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2374, + "step": 4940 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 4941 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 4942 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1644, + "step": 4943 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1678, + "step": 4944 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 4945 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1875, + "step": 4946 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1829, + "step": 4947 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.196, + "step": 4948 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2133, + "step": 4949 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2341, + "step": 4950 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1746, + "step": 4951 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1987, + "step": 4952 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1994, + "step": 4953 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 4954 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 4955 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1882, + "step": 4956 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 4957 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 4958 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 4959 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 4960 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1899, + "step": 4961 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1801, + "step": 4962 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1936, + "step": 4963 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1624, + "step": 4964 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1818, + "step": 4965 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2023, + "step": 4966 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1961, + "step": 4967 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1864, + "step": 4968 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1939, + "step": 4969 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1644, + "step": 4970 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2063, + "step": 4971 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1875, + "step": 4972 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1627, + "step": 4973 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1846, + "step": 4974 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.173, + "step": 4975 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1878, + "step": 4976 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 4977 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1975, + "step": 4978 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2108, + "step": 4979 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2004, + "step": 4980 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1664, + "step": 4981 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2058, + "step": 4982 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.174, + "step": 4983 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2171, + "step": 4984 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2296, + "step": 4985 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 4986 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1815, + "step": 4987 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 4988 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2013, + "step": 4989 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1988, + "step": 4990 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2536, + "step": 4991 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1801, + "step": 4992 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1976, + "step": 4993 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2162, + "step": 4994 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1902, + "step": 4995 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2226, + "step": 4996 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2202, + "step": 4997 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1534, + "step": 4998 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 4999 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1922, + "step": 5000 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1933, + "step": 5001 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1875, + "step": 5002 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1787, + "step": 5003 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2572, + "step": 5004 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2282, + "step": 5005 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1635, + "step": 5006 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1729, + "step": 5007 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1497, + "step": 5008 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1939, + "step": 5009 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2297, + "step": 5010 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1803, + "step": 5011 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 5012 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1598, + "step": 5013 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2078, + "step": 5014 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1918, + "step": 5015 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.1799, + "step": 5016 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.194, + "step": 5017 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2115, + "step": 5018 + }, + { + "epoch": 0.3, + "learning_rate": 2e-05, + "loss": 0.2531, + "step": 5019 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1692, + "step": 5020 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 5021 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 5022 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.2012, + "step": 5023 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1902, + "step": 5024 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1974, + "step": 5025 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1725, + "step": 5026 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.173, + "step": 5027 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.192, + "step": 5028 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.187, + "step": 5029 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1856, + "step": 5030 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1673, + "step": 5031 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1612, + "step": 5032 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1524, + "step": 5033 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.177, + "step": 5034 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.2016, + "step": 5035 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.168, + "step": 5036 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.2261, + "step": 5037 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 5038 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1705, + "step": 5039 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1734, + "step": 5040 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.175, + "step": 5041 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.21, + "step": 5042 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1513, + "step": 5043 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1761, + "step": 5044 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1647, + "step": 5045 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1714, + "step": 5046 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1586, + "step": 5047 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.2177, + "step": 5048 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1695, + "step": 5049 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1719, + "step": 5050 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1653, + "step": 5051 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1599, + "step": 5052 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.215, + "step": 5053 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1756, + "step": 5054 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1639, + "step": 5055 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1732, + "step": 5056 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1801, + "step": 5057 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.175, + "step": 5058 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1965, + "step": 5059 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1743, + "step": 5060 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1897, + "step": 5061 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1891, + "step": 5062 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.2305, + "step": 5063 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1713, + "step": 5064 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 5065 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1723, + "step": 5066 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1752, + "step": 5067 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.2435, + "step": 5068 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.2284, + "step": 5069 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 5070 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1734, + "step": 5071 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.198, + "step": 5072 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.2043, + "step": 5073 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.191, + "step": 5074 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1528, + "step": 5075 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1992, + "step": 5076 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1857, + "step": 5077 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1786, + "step": 5078 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1585, + "step": 5079 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1917, + "step": 5080 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1893, + "step": 5081 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.159, + "step": 5082 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1909, + "step": 5083 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 5084 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.2461, + "step": 5085 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 5086 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1909, + "step": 5087 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 5088 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 5089 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.2048, + "step": 5090 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.2248, + "step": 5091 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1525, + "step": 5092 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1524, + "step": 5093 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1848, + "step": 5094 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1809, + "step": 5095 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.2075, + "step": 5096 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1891, + "step": 5097 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 5098 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1804, + "step": 5099 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.2145, + "step": 5100 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1818, + "step": 5101 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1914, + "step": 5102 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.2076, + "step": 5103 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.2493, + "step": 5104 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1848, + "step": 5105 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1954, + "step": 5106 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1577, + "step": 5107 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.192, + "step": 5108 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1657, + "step": 5109 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.2405, + "step": 5110 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.2043, + "step": 5111 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1624, + "step": 5112 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1799, + "step": 5113 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.194, + "step": 5114 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.2262, + "step": 5115 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1936, + "step": 5116 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1692, + "step": 5117 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.175, + "step": 5118 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1981, + "step": 5119 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1652, + "step": 5120 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1964, + "step": 5121 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1713, + "step": 5122 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1857, + "step": 5123 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.2556, + "step": 5124 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 5125 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 5126 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.2399, + "step": 5127 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1949, + "step": 5128 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.2051, + "step": 5129 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1879, + "step": 5130 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1596, + "step": 5131 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1844, + "step": 5132 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1915, + "step": 5133 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 5134 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.2335, + "step": 5135 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.2087, + "step": 5136 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1782, + "step": 5137 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1872, + "step": 5138 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1771, + "step": 5139 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1875, + "step": 5140 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.226, + "step": 5141 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1839, + "step": 5142 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.182, + "step": 5143 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1847, + "step": 5144 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1611, + "step": 5145 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.2205, + "step": 5146 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.189, + "step": 5147 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1831, + "step": 5148 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.2008, + "step": 5149 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1591, + "step": 5150 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.2015, + "step": 5151 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1631, + "step": 5152 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.2041, + "step": 5153 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 5154 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.2131, + "step": 5155 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1588, + "step": 5156 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1986, + "step": 5157 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1873, + "step": 5158 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1901, + "step": 5159 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.2304, + "step": 5160 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1882, + "step": 5161 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 5162 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1908, + "step": 5163 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1973, + "step": 5164 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.2097, + "step": 5165 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.2231, + "step": 5166 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1866, + "step": 5167 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1797, + "step": 5168 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.213, + "step": 5169 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1465, + "step": 5170 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 5171 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.2003, + "step": 5172 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 5173 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.2267, + "step": 5174 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.216, + "step": 5175 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1962, + "step": 5176 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1672, + "step": 5177 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 5178 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.2057, + "step": 5179 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1763, + "step": 5180 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.2361, + "step": 5181 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.209, + "step": 5182 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.1967, + "step": 5183 + }, + { + "epoch": 0.31, + "learning_rate": 2e-05, + "loss": 0.2117, + "step": 5184 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.175, + "step": 5185 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1654, + "step": 5186 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1825, + "step": 5187 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.185, + "step": 5188 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1739, + "step": 5189 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.2046, + "step": 5190 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1869, + "step": 5191 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.192, + "step": 5192 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1714, + "step": 5193 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1698, + "step": 5194 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1526, + "step": 5195 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1652, + "step": 5196 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1597, + "step": 5197 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1965, + "step": 5198 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1717, + "step": 5199 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1674, + "step": 5200 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1931, + "step": 5201 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1541, + "step": 5202 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.2226, + "step": 5203 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.225, + "step": 5204 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.2367, + "step": 5205 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1869, + "step": 5206 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 5207 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1553, + "step": 5208 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1593, + "step": 5209 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1946, + "step": 5210 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.2253, + "step": 5211 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1861, + "step": 5212 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.2018, + "step": 5213 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1607, + "step": 5214 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.2207, + "step": 5215 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.2, + "step": 5216 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1949, + "step": 5217 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1975, + "step": 5218 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 5219 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1598, + "step": 5220 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.2181, + "step": 5221 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.2004, + "step": 5222 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1567, + "step": 5223 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.2377, + "step": 5224 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1801, + "step": 5225 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1913, + "step": 5226 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 5227 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1742, + "step": 5228 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1677, + "step": 5229 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1812, + "step": 5230 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.226, + "step": 5231 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1919, + "step": 5232 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.2044, + "step": 5233 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.169, + "step": 5234 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1865, + "step": 5235 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1777, + "step": 5236 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 5237 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.2064, + "step": 5238 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1819, + "step": 5239 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1768, + "step": 5240 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1819, + "step": 5241 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1708, + "step": 5242 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 5243 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 5244 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.2335, + "step": 5245 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1753, + "step": 5246 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.172, + "step": 5247 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1774, + "step": 5248 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1754, + "step": 5249 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1856, + "step": 5250 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.2003, + "step": 5251 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1618, + "step": 5252 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1746, + "step": 5253 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.2018, + "step": 5254 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.2152, + "step": 5255 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.2148, + "step": 5256 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1958, + "step": 5257 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1758, + "step": 5258 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1722, + "step": 5259 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.2182, + "step": 5260 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.2187, + "step": 5261 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.2448, + "step": 5262 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1561, + "step": 5263 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1951, + "step": 5264 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1586, + "step": 5265 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1666, + "step": 5266 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1747, + "step": 5267 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1909, + "step": 5268 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.2049, + "step": 5269 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.193, + "step": 5270 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.2099, + "step": 5271 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.2192, + "step": 5272 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1526, + "step": 5273 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 5274 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1637, + "step": 5275 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.2143, + "step": 5276 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.2074, + "step": 5277 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.2097, + "step": 5278 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.2656, + "step": 5279 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1616, + "step": 5280 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1602, + "step": 5281 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1791, + "step": 5282 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.2484, + "step": 5283 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1756, + "step": 5284 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 5285 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 5286 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1799, + "step": 5287 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1909, + "step": 5288 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1874, + "step": 5289 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1857, + "step": 5290 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.2097, + "step": 5291 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1792, + "step": 5292 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 5293 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1682, + "step": 5294 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1913, + "step": 5295 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.2193, + "step": 5296 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1583, + "step": 5297 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 5298 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 5299 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1815, + "step": 5300 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1696, + "step": 5301 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 5302 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1723, + "step": 5303 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1795, + "step": 5304 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1836, + "step": 5305 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1945, + "step": 5306 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.2082, + "step": 5307 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1875, + "step": 5308 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 5309 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1897, + "step": 5310 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 5311 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.2454, + "step": 5312 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1892, + "step": 5313 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 5314 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1772, + "step": 5315 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.2372, + "step": 5316 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.2055, + "step": 5317 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1844, + "step": 5318 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 5319 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.22, + "step": 5320 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1724, + "step": 5321 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.2361, + "step": 5322 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1965, + "step": 5323 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1652, + "step": 5324 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.187, + "step": 5325 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 5326 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.2088, + "step": 5327 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1956, + "step": 5328 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.2145, + "step": 5329 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1649, + "step": 5330 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 5331 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1731, + "step": 5332 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1647, + "step": 5333 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1538, + "step": 5334 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.205, + "step": 5335 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1673, + "step": 5336 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1793, + "step": 5337 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.188, + "step": 5338 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 5339 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1524, + "step": 5340 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.2126, + "step": 5341 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1749, + "step": 5342 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1558, + "step": 5343 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1907, + "step": 5344 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.2464, + "step": 5345 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1772, + "step": 5346 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1718, + "step": 5347 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 5348 + }, + { + "epoch": 0.32, + "learning_rate": 2e-05, + "loss": 0.1803, + "step": 5349 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1563, + "step": 5350 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1528, + "step": 5351 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.2075, + "step": 5352 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.2173, + "step": 5353 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1934, + "step": 5354 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.2047, + "step": 5355 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1762, + "step": 5356 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1845, + "step": 5357 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.2098, + "step": 5358 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1704, + "step": 5359 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1692, + "step": 5360 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1895, + "step": 5361 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 5362 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1996, + "step": 5363 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.201, + "step": 5364 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1725, + "step": 5365 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.204, + "step": 5366 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.2386, + "step": 5367 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1899, + "step": 5368 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1647, + "step": 5369 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1629, + "step": 5370 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1867, + "step": 5371 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.2023, + "step": 5372 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1564, + "step": 5373 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1702, + "step": 5374 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.215, + "step": 5375 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 5376 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1838, + "step": 5377 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1947, + "step": 5378 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 5379 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1933, + "step": 5380 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 5381 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.2013, + "step": 5382 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1753, + "step": 5383 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1856, + "step": 5384 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 5385 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1783, + "step": 5386 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 5387 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1978, + "step": 5388 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.2089, + "step": 5389 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.195, + "step": 5390 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.174, + "step": 5391 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.2028, + "step": 5392 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.2173, + "step": 5393 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1788, + "step": 5394 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1891, + "step": 5395 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1624, + "step": 5396 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.171, + "step": 5397 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.186, + "step": 5398 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1831, + "step": 5399 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1998, + "step": 5400 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1819, + "step": 5401 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1847, + "step": 5402 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.265, + "step": 5403 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1985, + "step": 5404 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.184, + "step": 5405 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.146, + "step": 5406 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1583, + "step": 5407 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1639, + "step": 5408 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 5409 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1634, + "step": 5410 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1852, + "step": 5411 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1488, + "step": 5412 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1763, + "step": 5413 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1852, + "step": 5414 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 5415 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.2031, + "step": 5416 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1778, + "step": 5417 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1462, + "step": 5418 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.2088, + "step": 5419 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1982, + "step": 5420 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1748, + "step": 5421 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1974, + "step": 5422 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1668, + "step": 5423 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.2266, + "step": 5424 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1652, + "step": 5425 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1827, + "step": 5426 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 5427 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 5428 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.2044, + "step": 5429 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.2168, + "step": 5430 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.2002, + "step": 5431 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1644, + "step": 5432 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 5433 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.2046, + "step": 5434 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1836, + "step": 5435 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1906, + "step": 5436 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 5437 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1728, + "step": 5438 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1853, + "step": 5439 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1804, + "step": 5440 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1554, + "step": 5441 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.155, + "step": 5442 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1821, + "step": 5443 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.187, + "step": 5444 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1564, + "step": 5445 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.198, + "step": 5446 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.2168, + "step": 5447 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 5448 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.2446, + "step": 5449 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1996, + "step": 5450 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1633, + "step": 5451 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1977, + "step": 5452 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.201, + "step": 5453 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.2097, + "step": 5454 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1625, + "step": 5455 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 5456 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 5457 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.2351, + "step": 5458 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 5459 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.2161, + "step": 5460 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 5461 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1759, + "step": 5462 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1806, + "step": 5463 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1746, + "step": 5464 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1689, + "step": 5465 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.197, + "step": 5466 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1607, + "step": 5467 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1807, + "step": 5468 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.2152, + "step": 5469 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.172, + "step": 5470 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1786, + "step": 5471 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1879, + "step": 5472 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 5473 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1693, + "step": 5474 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 5475 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.2442, + "step": 5476 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1698, + "step": 5477 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 5478 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1823, + "step": 5479 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.2069, + "step": 5480 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1884, + "step": 5481 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1608, + "step": 5482 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.2088, + "step": 5483 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1986, + "step": 5484 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1645, + "step": 5485 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 5486 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 5487 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 5488 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 5489 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.2122, + "step": 5490 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1863, + "step": 5491 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1993, + "step": 5492 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.2192, + "step": 5493 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.2311, + "step": 5494 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 5495 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.21, + "step": 5496 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1743, + "step": 5497 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1621, + "step": 5498 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.2126, + "step": 5499 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1937, + "step": 5500 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1552, + "step": 5501 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.161, + "step": 5502 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1803, + "step": 5503 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1633, + "step": 5504 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1966, + "step": 5505 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.2167, + "step": 5506 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1749, + "step": 5507 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1591, + "step": 5508 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1469, + "step": 5509 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.241, + "step": 5510 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.2415, + "step": 5511 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 5512 + }, + { + "epoch": 0.33, + "learning_rate": 2e-05, + "loss": 0.1606, + "step": 5513 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1773, + "step": 5514 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1913, + "step": 5515 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1742, + "step": 5516 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1803, + "step": 5517 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1766, + "step": 5518 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1877, + "step": 5519 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 5520 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.209, + "step": 5521 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1671, + "step": 5522 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.209, + "step": 5523 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.146, + "step": 5524 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.162, + "step": 5525 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 5526 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1665, + "step": 5527 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 5528 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.2134, + "step": 5529 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.195, + "step": 5530 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.2084, + "step": 5531 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1788, + "step": 5532 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1841, + "step": 5533 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 5534 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1808, + "step": 5535 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.2344, + "step": 5536 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1886, + "step": 5537 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1786, + "step": 5538 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 5539 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1686, + "step": 5540 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.2048, + "step": 5541 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.169, + "step": 5542 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1632, + "step": 5543 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.2108, + "step": 5544 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.2396, + "step": 5545 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1772, + "step": 5546 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1877, + "step": 5547 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.2051, + "step": 5548 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1952, + "step": 5549 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1838, + "step": 5550 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1858, + "step": 5551 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1987, + "step": 5552 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 5553 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.2253, + "step": 5554 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.2107, + "step": 5555 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1781, + "step": 5556 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1937, + "step": 5557 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1844, + "step": 5558 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1807, + "step": 5559 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.2033, + "step": 5560 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1901, + "step": 5561 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1807, + "step": 5562 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 5563 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.2006, + "step": 5564 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 5565 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1789, + "step": 5566 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1751, + "step": 5567 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1566, + "step": 5568 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1826, + "step": 5569 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1628, + "step": 5570 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1771, + "step": 5571 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1707, + "step": 5572 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1811, + "step": 5573 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1756, + "step": 5574 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1881, + "step": 5575 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1616, + "step": 5576 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1595, + "step": 5577 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.208, + "step": 5578 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.2063, + "step": 5579 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.2255, + "step": 5580 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1925, + "step": 5581 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.2272, + "step": 5582 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1634, + "step": 5583 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 5584 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 5585 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1456, + "step": 5586 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.2137, + "step": 5587 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1613, + "step": 5588 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1917, + "step": 5589 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1468, + "step": 5590 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 5591 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.2209, + "step": 5592 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1659, + "step": 5593 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1714, + "step": 5594 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1847, + "step": 5595 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 5596 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1583, + "step": 5597 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1808, + "step": 5598 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1558, + "step": 5599 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 5600 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1634, + "step": 5601 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1662, + "step": 5602 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1994, + "step": 5603 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1747, + "step": 5604 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1628, + "step": 5605 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1768, + "step": 5606 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1747, + "step": 5607 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1721, + "step": 5608 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.224, + "step": 5609 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1583, + "step": 5610 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 5611 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1634, + "step": 5612 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1456, + "step": 5613 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.2118, + "step": 5614 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 5615 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.2069, + "step": 5616 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.2344, + "step": 5617 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1562, + "step": 5618 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1774, + "step": 5619 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.2118, + "step": 5620 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1838, + "step": 5621 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1875, + "step": 5622 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.203, + "step": 5623 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1618, + "step": 5624 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 5625 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 5626 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 5627 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1933, + "step": 5628 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1704, + "step": 5629 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.2137, + "step": 5630 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.172, + "step": 5631 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.2351, + "step": 5632 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1982, + "step": 5633 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.185, + "step": 5634 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1813, + "step": 5635 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1831, + "step": 5636 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1646, + "step": 5637 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.216, + "step": 5638 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 5639 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.2905, + "step": 5640 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.192, + "step": 5641 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1767, + "step": 5642 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1761, + "step": 5643 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1712, + "step": 5644 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1757, + "step": 5645 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1658, + "step": 5646 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1638, + "step": 5647 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1919, + "step": 5648 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.175, + "step": 5649 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1811, + "step": 5650 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.2222, + "step": 5651 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1687, + "step": 5652 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.2175, + "step": 5653 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.2089, + "step": 5654 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.2257, + "step": 5655 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1831, + "step": 5656 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1876, + "step": 5657 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1777, + "step": 5658 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.2148, + "step": 5659 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1952, + "step": 5660 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.2345, + "step": 5661 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1667, + "step": 5662 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.2151, + "step": 5663 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.2031, + "step": 5664 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1596, + "step": 5665 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.2061, + "step": 5666 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 5667 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1687, + "step": 5668 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1933, + "step": 5669 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1932, + "step": 5670 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1812, + "step": 5671 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1997, + "step": 5672 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.2061, + "step": 5673 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1665, + "step": 5674 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1842, + "step": 5675 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1621, + "step": 5676 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1883, + "step": 5677 + }, + { + "epoch": 0.34, + "learning_rate": 2e-05, + "loss": 0.1617, + "step": 5678 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1965, + "step": 5679 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1546, + "step": 5680 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 5681 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1762, + "step": 5682 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1442, + "step": 5683 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1773, + "step": 5684 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.179, + "step": 5685 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1977, + "step": 5686 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1653, + "step": 5687 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.2062, + "step": 5688 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.2898, + "step": 5689 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1753, + "step": 5690 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.2163, + "step": 5691 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1628, + "step": 5692 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1963, + "step": 5693 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1652, + "step": 5694 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.2299, + "step": 5695 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1844, + "step": 5696 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 5697 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1851, + "step": 5698 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1769, + "step": 5699 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1621, + "step": 5700 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1882, + "step": 5701 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 5702 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 5703 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1655, + "step": 5704 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1919, + "step": 5705 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1693, + "step": 5706 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1822, + "step": 5707 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.2033, + "step": 5708 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1994, + "step": 5709 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1819, + "step": 5710 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.2102, + "step": 5711 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1589, + "step": 5712 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.2342, + "step": 5713 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.2534, + "step": 5714 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1659, + "step": 5715 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.2353, + "step": 5716 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1599, + "step": 5717 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.2308, + "step": 5718 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 5719 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1462, + "step": 5720 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 5721 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1967, + "step": 5722 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1582, + "step": 5723 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 5724 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.2007, + "step": 5725 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1936, + "step": 5726 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1721, + "step": 5727 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.2219, + "step": 5728 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1632, + "step": 5729 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1743, + "step": 5730 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 5731 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1598, + "step": 5732 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 5733 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.181, + "step": 5734 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1904, + "step": 5735 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1907, + "step": 5736 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1804, + "step": 5737 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1674, + "step": 5738 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.165, + "step": 5739 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1936, + "step": 5740 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.2187, + "step": 5741 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.2322, + "step": 5742 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.2289, + "step": 5743 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1726, + "step": 5744 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 5745 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1899, + "step": 5746 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1672, + "step": 5747 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1917, + "step": 5748 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1686, + "step": 5749 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1692, + "step": 5750 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.193, + "step": 5751 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1943, + "step": 5752 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1777, + "step": 5753 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1869, + "step": 5754 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1592, + "step": 5755 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1638, + "step": 5756 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1834, + "step": 5757 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1633, + "step": 5758 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1957, + "step": 5759 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.2303, + "step": 5760 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1843, + "step": 5761 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 5762 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.2063, + "step": 5763 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1906, + "step": 5764 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 5765 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1599, + "step": 5766 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1897, + "step": 5767 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.2114, + "step": 5768 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.2472, + "step": 5769 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1912, + "step": 5770 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1485, + "step": 5771 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1854, + "step": 5772 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 5773 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 5774 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1525, + "step": 5775 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1735, + "step": 5776 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 5777 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1953, + "step": 5778 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1902, + "step": 5779 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.2245, + "step": 5780 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.2237, + "step": 5781 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 5782 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1596, + "step": 5783 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.2213, + "step": 5784 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1719, + "step": 5785 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1586, + "step": 5786 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.2102, + "step": 5787 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1716, + "step": 5788 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1942, + "step": 5789 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1701, + "step": 5790 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1747, + "step": 5791 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1924, + "step": 5792 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.2203, + "step": 5793 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1906, + "step": 5794 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1724, + "step": 5795 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1592, + "step": 5796 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.2565, + "step": 5797 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1702, + "step": 5798 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.2005, + "step": 5799 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1827, + "step": 5800 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1797, + "step": 5801 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 5802 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1573, + "step": 5803 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1713, + "step": 5804 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.165, + "step": 5805 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1612, + "step": 5806 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1936, + "step": 5807 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.199, + "step": 5808 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.203, + "step": 5809 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 5810 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1928, + "step": 5811 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.2157, + "step": 5812 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1949, + "step": 5813 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1681, + "step": 5814 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.2378, + "step": 5815 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1994, + "step": 5816 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1893, + "step": 5817 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1564, + "step": 5818 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.2094, + "step": 5819 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1851, + "step": 5820 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 5821 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1525, + "step": 5822 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1638, + "step": 5823 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.2296, + "step": 5824 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.2032, + "step": 5825 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 5826 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.2226, + "step": 5827 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.2353, + "step": 5828 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.2063, + "step": 5829 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.2016, + "step": 5830 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.2057, + "step": 5831 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1657, + "step": 5832 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1926, + "step": 5833 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.2042, + "step": 5834 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1758, + "step": 5835 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.2336, + "step": 5836 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1525, + "step": 5837 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 5838 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.2418, + "step": 5839 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1752, + "step": 5840 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1649, + "step": 5841 + }, + { + "epoch": 0.35, + "learning_rate": 2e-05, + "loss": 0.1869, + "step": 5842 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.2137, + "step": 5843 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1649, + "step": 5844 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1826, + "step": 5845 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.192, + "step": 5846 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 5847 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.191, + "step": 5848 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1903, + "step": 5849 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 5850 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1676, + "step": 5851 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.199, + "step": 5852 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1919, + "step": 5853 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1666, + "step": 5854 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1663, + "step": 5855 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1668, + "step": 5856 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 5857 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.2076, + "step": 5858 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 5859 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.2125, + "step": 5860 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1639, + "step": 5861 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.2244, + "step": 5862 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.201, + "step": 5863 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1614, + "step": 5864 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1768, + "step": 5865 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.171, + "step": 5866 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1622, + "step": 5867 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1716, + "step": 5868 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.212, + "step": 5869 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1787, + "step": 5870 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.2415, + "step": 5871 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.153, + "step": 5872 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1598, + "step": 5873 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1616, + "step": 5874 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.2181, + "step": 5875 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 5876 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 5877 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1725, + "step": 5878 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.175, + "step": 5879 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.2176, + "step": 5880 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.2661, + "step": 5881 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1638, + "step": 5882 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.237, + "step": 5883 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1606, + "step": 5884 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1447, + "step": 5885 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.2244, + "step": 5886 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 5887 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1826, + "step": 5888 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.154, + "step": 5889 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 5890 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1977, + "step": 5891 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1742, + "step": 5892 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.227, + "step": 5893 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.147, + "step": 5894 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.144, + "step": 5895 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1695, + "step": 5896 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1785, + "step": 5897 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.2127, + "step": 5898 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1792, + "step": 5899 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.181, + "step": 5900 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 5901 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 5902 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1841, + "step": 5903 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.2217, + "step": 5904 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.2162, + "step": 5905 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.2093, + "step": 5906 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.2255, + "step": 5907 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1824, + "step": 5908 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 5909 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1611, + "step": 5910 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 5911 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.2225, + "step": 5912 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1834, + "step": 5913 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.2063, + "step": 5914 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1595, + "step": 5915 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.2129, + "step": 5916 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1871, + "step": 5917 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.145, + "step": 5918 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1764, + "step": 5919 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.2015, + "step": 5920 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 5921 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1761, + "step": 5922 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1831, + "step": 5923 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1746, + "step": 5924 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.2077, + "step": 5925 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1725, + "step": 5926 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1734, + "step": 5927 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1922, + "step": 5928 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.2047, + "step": 5929 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.218, + "step": 5930 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 5931 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1909, + "step": 5932 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.2155, + "step": 5933 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1814, + "step": 5934 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1892, + "step": 5935 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1946, + "step": 5936 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1695, + "step": 5937 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 5938 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1727, + "step": 5939 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1751, + "step": 5940 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1979, + "step": 5941 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.255, + "step": 5942 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1747, + "step": 5943 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 5944 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1704, + "step": 5945 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1865, + "step": 5946 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.2173, + "step": 5947 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.2027, + "step": 5948 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.199, + "step": 5949 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.2047, + "step": 5950 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.2057, + "step": 5951 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1643, + "step": 5952 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1674, + "step": 5953 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.214, + "step": 5954 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1952, + "step": 5955 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.2308, + "step": 5956 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1657, + "step": 5957 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.2, + "step": 5958 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1775, + "step": 5959 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1608, + "step": 5960 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1511, + "step": 5961 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 5962 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.188, + "step": 5963 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 5964 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1745, + "step": 5965 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 5966 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1865, + "step": 5967 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1603, + "step": 5968 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.194, + "step": 5969 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1715, + "step": 5970 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1997, + "step": 5971 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1724, + "step": 5972 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1561, + "step": 5973 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.2036, + "step": 5974 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1682, + "step": 5975 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 5976 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 5977 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1738, + "step": 5978 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1698, + "step": 5979 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 5980 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1677, + "step": 5981 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1683, + "step": 5982 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.2346, + "step": 5983 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.172, + "step": 5984 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 5985 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.2277, + "step": 5986 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.183, + "step": 5987 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.2157, + "step": 5988 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1902, + "step": 5989 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1752, + "step": 5990 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.163, + "step": 5991 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 5992 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 5993 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1468, + "step": 5994 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1713, + "step": 5995 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1929, + "step": 5996 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1658, + "step": 5997 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1787, + "step": 5998 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 5999 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1476, + "step": 6000 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1952, + "step": 6001 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1891, + "step": 6002 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.2227, + "step": 6003 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1589, + "step": 6004 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1488, + "step": 6005 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.1809, + "step": 6006 + }, + { + "epoch": 0.36, + "learning_rate": 2e-05, + "loss": 0.2168, + "step": 6007 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.22, + "step": 6008 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1515, + "step": 6009 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2177, + "step": 6010 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1677, + "step": 6011 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2411, + "step": 6012 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2344, + "step": 6013 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1602, + "step": 6014 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.203, + "step": 6015 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1867, + "step": 6016 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2072, + "step": 6017 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1864, + "step": 6018 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 6019 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1917, + "step": 6020 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.161, + "step": 6021 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2259, + "step": 6022 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 6023 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1812, + "step": 6024 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 6025 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1793, + "step": 6026 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1963, + "step": 6027 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1521, + "step": 6028 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1632, + "step": 6029 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 6030 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1447, + "step": 6031 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1669, + "step": 6032 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.177, + "step": 6033 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1859, + "step": 6034 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2083, + "step": 6035 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.19, + "step": 6036 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.244, + "step": 6037 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1743, + "step": 6038 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1893, + "step": 6039 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.172, + "step": 6040 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 6041 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1712, + "step": 6042 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2287, + "step": 6043 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2162, + "step": 6044 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1831, + "step": 6045 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1913, + "step": 6046 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1713, + "step": 6047 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2026, + "step": 6048 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1878, + "step": 6049 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1906, + "step": 6050 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1667, + "step": 6051 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.199, + "step": 6052 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2312, + "step": 6053 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1681, + "step": 6054 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1478, + "step": 6055 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1673, + "step": 6056 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1802, + "step": 6057 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1829, + "step": 6058 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1916, + "step": 6059 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 6060 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1691, + "step": 6061 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2418, + "step": 6062 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1983, + "step": 6063 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2187, + "step": 6064 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1822, + "step": 6065 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2164, + "step": 6066 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.167, + "step": 6067 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2185, + "step": 6068 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.209, + "step": 6069 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2095, + "step": 6070 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1677, + "step": 6071 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1604, + "step": 6072 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1706, + "step": 6073 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2022, + "step": 6074 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.153, + "step": 6075 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1495, + "step": 6076 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1888, + "step": 6077 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2044, + "step": 6078 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2271, + "step": 6079 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2081, + "step": 6080 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1577, + "step": 6081 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1586, + "step": 6082 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2111, + "step": 6083 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1813, + "step": 6084 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 6085 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2227, + "step": 6086 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.176, + "step": 6087 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2105, + "step": 6088 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1884, + "step": 6089 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1772, + "step": 6090 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2082, + "step": 6091 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2206, + "step": 6092 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1531, + "step": 6093 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2019, + "step": 6094 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1722, + "step": 6095 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2338, + "step": 6096 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 6097 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1548, + "step": 6098 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.176, + "step": 6099 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1698, + "step": 6100 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.147, + "step": 6101 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 6102 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1691, + "step": 6103 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2357, + "step": 6104 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1705, + "step": 6105 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1993, + "step": 6106 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2098, + "step": 6107 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2024, + "step": 6108 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2124, + "step": 6109 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1676, + "step": 6110 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1801, + "step": 6111 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1982, + "step": 6112 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.188, + "step": 6113 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2069, + "step": 6114 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1908, + "step": 6115 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2143, + "step": 6116 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1814, + "step": 6117 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1861, + "step": 6118 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1771, + "step": 6119 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1887, + "step": 6120 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.194, + "step": 6121 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1909, + "step": 6122 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2552, + "step": 6123 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2121, + "step": 6124 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2075, + "step": 6125 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2048, + "step": 6126 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2157, + "step": 6127 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2069, + "step": 6128 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 6129 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1732, + "step": 6130 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1784, + "step": 6131 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 6132 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1576, + "step": 6133 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1814, + "step": 6134 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2058, + "step": 6135 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1686, + "step": 6136 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1865, + "step": 6137 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 6138 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1535, + "step": 6139 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1939, + "step": 6140 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2018, + "step": 6141 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1527, + "step": 6142 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1842, + "step": 6143 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2668, + "step": 6144 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1688, + "step": 6145 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1848, + "step": 6146 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1725, + "step": 6147 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2082, + "step": 6148 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1633, + "step": 6149 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1688, + "step": 6150 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 6151 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.222, + "step": 6152 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2191, + "step": 6153 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1576, + "step": 6154 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2108, + "step": 6155 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 6156 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.2625, + "step": 6157 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1733, + "step": 6158 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1864, + "step": 6159 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1625, + "step": 6160 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 6161 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 6162 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1793, + "step": 6163 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1513, + "step": 6164 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1735, + "step": 6165 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1882, + "step": 6166 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 6167 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1739, + "step": 6168 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1773, + "step": 6169 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1857, + "step": 6170 + }, + { + "epoch": 0.37, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 6171 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 6172 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.2108, + "step": 6173 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1926, + "step": 6174 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.2026, + "step": 6175 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1725, + "step": 6176 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1861, + "step": 6177 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 6178 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1548, + "step": 6179 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1443, + "step": 6180 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1874, + "step": 6181 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1674, + "step": 6182 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.2398, + "step": 6183 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1652, + "step": 6184 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1886, + "step": 6185 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1511, + "step": 6186 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1783, + "step": 6187 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1757, + "step": 6188 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1936, + "step": 6189 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1828, + "step": 6190 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 6191 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.2473, + "step": 6192 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1803, + "step": 6193 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1777, + "step": 6194 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1488, + "step": 6195 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1638, + "step": 6196 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 6197 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.2011, + "step": 6198 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 6199 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1835, + "step": 6200 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 6201 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1707, + "step": 6202 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1601, + "step": 6203 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.2091, + "step": 6204 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1639, + "step": 6205 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.2055, + "step": 6206 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 6207 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 6208 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1646, + "step": 6209 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1631, + "step": 6210 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 6211 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1797, + "step": 6212 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.176, + "step": 6213 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1596, + "step": 6214 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1549, + "step": 6215 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.2163, + "step": 6216 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1987, + "step": 6217 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.2053, + "step": 6218 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.2016, + "step": 6219 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 6220 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.2073, + "step": 6221 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1813, + "step": 6222 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1618, + "step": 6223 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1748, + "step": 6224 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.2078, + "step": 6225 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.2079, + "step": 6226 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1925, + "step": 6227 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1863, + "step": 6228 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.161, + "step": 6229 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 6230 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1755, + "step": 6231 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1763, + "step": 6232 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1975, + "step": 6233 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.2004, + "step": 6234 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1738, + "step": 6235 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.2109, + "step": 6236 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1576, + "step": 6237 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1964, + "step": 6238 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1637, + "step": 6239 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1719, + "step": 6240 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1668, + "step": 6241 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1622, + "step": 6242 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 6243 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1845, + "step": 6244 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.2355, + "step": 6245 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.2101, + "step": 6246 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1797, + "step": 6247 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1714, + "step": 6248 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1769, + "step": 6249 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1849, + "step": 6250 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1715, + "step": 6251 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1937, + "step": 6252 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1613, + "step": 6253 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.2029, + "step": 6254 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.2037, + "step": 6255 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.2384, + "step": 6256 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.2064, + "step": 6257 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1764, + "step": 6258 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1627, + "step": 6259 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1718, + "step": 6260 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.2091, + "step": 6261 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1775, + "step": 6262 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.2246, + "step": 6263 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1994, + "step": 6264 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.257, + "step": 6265 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.2213, + "step": 6266 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1666, + "step": 6267 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1933, + "step": 6268 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.2228, + "step": 6269 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.164, + "step": 6270 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.155, + "step": 6271 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.2163, + "step": 6272 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1778, + "step": 6273 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1591, + "step": 6274 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1798, + "step": 6275 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1937, + "step": 6276 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.2138, + "step": 6277 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.2194, + "step": 6278 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.2061, + "step": 6279 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1798, + "step": 6280 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 6281 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 6282 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.171, + "step": 6283 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.236, + "step": 6284 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1862, + "step": 6285 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1683, + "step": 6286 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1797, + "step": 6287 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.2153, + "step": 6288 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.191, + "step": 6289 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1669, + "step": 6290 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1912, + "step": 6291 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.154, + "step": 6292 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1675, + "step": 6293 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1721, + "step": 6294 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.2123, + "step": 6295 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1853, + "step": 6296 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.162, + "step": 6297 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 6298 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1975, + "step": 6299 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1781, + "step": 6300 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1788, + "step": 6301 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1642, + "step": 6302 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1988, + "step": 6303 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1685, + "step": 6304 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.2059, + "step": 6305 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.2292, + "step": 6306 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 6307 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1661, + "step": 6308 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1958, + "step": 6309 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.2005, + "step": 6310 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1442, + "step": 6311 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 6312 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1683, + "step": 6313 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 6314 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1932, + "step": 6315 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 6316 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1734, + "step": 6317 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.2038, + "step": 6318 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1808, + "step": 6319 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 6320 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1921, + "step": 6321 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.2033, + "step": 6322 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1826, + "step": 6323 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1874, + "step": 6324 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1876, + "step": 6325 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 6326 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1682, + "step": 6327 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1982, + "step": 6328 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1776, + "step": 6329 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1695, + "step": 6330 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1944, + "step": 6331 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1632, + "step": 6332 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1893, + "step": 6333 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.2016, + "step": 6334 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.195, + "step": 6335 + }, + { + "epoch": 0.38, + "learning_rate": 2e-05, + "loss": 0.1898, + "step": 6336 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.171, + "step": 6337 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1884, + "step": 6338 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.161, + "step": 6339 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.2153, + "step": 6340 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 6341 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1857, + "step": 6342 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1965, + "step": 6343 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1557, + "step": 6344 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.216, + "step": 6345 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.2008, + "step": 6346 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1686, + "step": 6347 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1837, + "step": 6348 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 6349 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1762, + "step": 6350 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 6351 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 6352 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1989, + "step": 6353 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1883, + "step": 6354 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.2016, + "step": 6355 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1601, + "step": 6356 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1624, + "step": 6357 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1826, + "step": 6358 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1781, + "step": 6359 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1788, + "step": 6360 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1876, + "step": 6361 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.2171, + "step": 6362 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1835, + "step": 6363 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.2368, + "step": 6364 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.2654, + "step": 6365 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.2075, + "step": 6366 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.2498, + "step": 6367 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.2156, + "step": 6368 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1845, + "step": 6369 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 6370 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.24, + "step": 6371 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1834, + "step": 6372 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1623, + "step": 6373 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1695, + "step": 6374 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1802, + "step": 6375 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1945, + "step": 6376 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.163, + "step": 6377 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.2053, + "step": 6378 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 6379 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 6380 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1456, + "step": 6381 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1924, + "step": 6382 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1888, + "step": 6383 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1824, + "step": 6384 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.2411, + "step": 6385 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1519, + "step": 6386 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1598, + "step": 6387 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1744, + "step": 6388 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 6389 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1739, + "step": 6390 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.185, + "step": 6391 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1956, + "step": 6392 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1993, + "step": 6393 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1889, + "step": 6394 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1996, + "step": 6395 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1657, + "step": 6396 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1772, + "step": 6397 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1645, + "step": 6398 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.2172, + "step": 6399 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 6400 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1502, + "step": 6401 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1884, + "step": 6402 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1497, + "step": 6403 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.169, + "step": 6404 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1884, + "step": 6405 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1948, + "step": 6406 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.2069, + "step": 6407 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1582, + "step": 6408 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 6409 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1929, + "step": 6410 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1841, + "step": 6411 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 6412 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 6413 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.189, + "step": 6414 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1585, + "step": 6415 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.2354, + "step": 6416 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1644, + "step": 6417 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1677, + "step": 6418 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1964, + "step": 6419 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1531, + "step": 6420 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 6421 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1651, + "step": 6422 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1893, + "step": 6423 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 6424 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.2131, + "step": 6425 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1972, + "step": 6426 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1877, + "step": 6427 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.2079, + "step": 6428 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.147, + "step": 6429 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1476, + "step": 6430 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 6431 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 6432 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1541, + "step": 6433 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1834, + "step": 6434 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1845, + "step": 6435 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1894, + "step": 6436 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1841, + "step": 6437 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1939, + "step": 6438 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1833, + "step": 6439 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1563, + "step": 6440 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1874, + "step": 6441 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1598, + "step": 6442 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 6443 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.2321, + "step": 6444 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.207, + "step": 6445 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1627, + "step": 6446 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1876, + "step": 6447 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1822, + "step": 6448 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 6449 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1836, + "step": 6450 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 6451 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 6452 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1851, + "step": 6453 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1845, + "step": 6454 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 6455 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1867, + "step": 6456 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1682, + "step": 6457 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1858, + "step": 6458 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1687, + "step": 6459 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1655, + "step": 6460 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.2043, + "step": 6461 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1818, + "step": 6462 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 6463 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1899, + "step": 6464 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 6465 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 6466 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.2144, + "step": 6467 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.2084, + "step": 6468 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.2045, + "step": 6469 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.2176, + "step": 6470 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.188, + "step": 6471 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1667, + "step": 6472 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1851, + "step": 6473 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1657, + "step": 6474 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1641, + "step": 6475 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.2424, + "step": 6476 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 6477 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1631, + "step": 6478 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1644, + "step": 6479 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 6480 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 6481 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1595, + "step": 6482 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 6483 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1715, + "step": 6484 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1635, + "step": 6485 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.2049, + "step": 6486 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1668, + "step": 6487 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1836, + "step": 6488 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1676, + "step": 6489 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1924, + "step": 6490 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1849, + "step": 6491 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1943, + "step": 6492 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 6493 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1828, + "step": 6494 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1886, + "step": 6495 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.2288, + "step": 6496 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 6497 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1691, + "step": 6498 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1688, + "step": 6499 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1776, + "step": 6500 + }, + { + "epoch": 0.39, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 6501 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.184, + "step": 6502 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1979, + "step": 6503 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.158, + "step": 6504 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1681, + "step": 6505 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.2172, + "step": 6506 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.2383, + "step": 6507 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1648, + "step": 6508 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.161, + "step": 6509 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.2198, + "step": 6510 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 6511 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.2203, + "step": 6512 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.2071, + "step": 6513 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1745, + "step": 6514 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.2017, + "step": 6515 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 6516 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 6517 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 6518 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1849, + "step": 6519 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1925, + "step": 6520 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.207, + "step": 6521 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1655, + "step": 6522 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 6523 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.2334, + "step": 6524 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.2425, + "step": 6525 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1908, + "step": 6526 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.153, + "step": 6527 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1616, + "step": 6528 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.2052, + "step": 6529 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.2048, + "step": 6530 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1952, + "step": 6531 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 6532 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1707, + "step": 6533 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.2287, + "step": 6534 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1892, + "step": 6535 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.154, + "step": 6536 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1861, + "step": 6537 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1867, + "step": 6538 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1644, + "step": 6539 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1977, + "step": 6540 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1642, + "step": 6541 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 6542 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 6543 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1538, + "step": 6544 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 6545 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1962, + "step": 6546 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1675, + "step": 6547 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.2117, + "step": 6548 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1632, + "step": 6549 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1462, + "step": 6550 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1756, + "step": 6551 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.156, + "step": 6552 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1773, + "step": 6553 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 6554 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1697, + "step": 6555 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.2052, + "step": 6556 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1567, + "step": 6557 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.169, + "step": 6558 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1762, + "step": 6559 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.2142, + "step": 6560 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1997, + "step": 6561 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 6562 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1813, + "step": 6563 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1549, + "step": 6564 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1534, + "step": 6565 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1902, + "step": 6566 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.2041, + "step": 6567 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 6568 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 6569 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1618, + "step": 6570 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 6571 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 6572 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.228, + "step": 6573 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.2188, + "step": 6574 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1777, + "step": 6575 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1909, + "step": 6576 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1626, + "step": 6577 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1681, + "step": 6578 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1851, + "step": 6579 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 6580 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.197, + "step": 6581 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1848, + "step": 6582 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 6583 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1663, + "step": 6584 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1776, + "step": 6585 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 6586 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.184, + "step": 6587 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1731, + "step": 6588 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 6589 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1927, + "step": 6590 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1903, + "step": 6591 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.2074, + "step": 6592 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 6593 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1751, + "step": 6594 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1855, + "step": 6595 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1608, + "step": 6596 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1764, + "step": 6597 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1775, + "step": 6598 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1807, + "step": 6599 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1668, + "step": 6600 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 6601 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1734, + "step": 6602 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1592, + "step": 6603 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1891, + "step": 6604 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 6605 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1813, + "step": 6606 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 6607 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1946, + "step": 6608 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1692, + "step": 6609 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.2149, + "step": 6610 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.182, + "step": 6611 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.2044, + "step": 6612 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1588, + "step": 6613 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.2039, + "step": 6614 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1895, + "step": 6615 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.2264, + "step": 6616 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.2215, + "step": 6617 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1882, + "step": 6618 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1756, + "step": 6619 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1857, + "step": 6620 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.214, + "step": 6621 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1842, + "step": 6622 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 6623 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1737, + "step": 6624 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1692, + "step": 6625 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1839, + "step": 6626 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1683, + "step": 6627 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1685, + "step": 6628 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1804, + "step": 6629 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.2058, + "step": 6630 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1611, + "step": 6631 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1757, + "step": 6632 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1927, + "step": 6633 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1462, + "step": 6634 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.184, + "step": 6635 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1617, + "step": 6636 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1588, + "step": 6637 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1946, + "step": 6638 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1726, + "step": 6639 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1944, + "step": 6640 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1538, + "step": 6641 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.2319, + "step": 6642 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1778, + "step": 6643 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.2089, + "step": 6644 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 6645 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1552, + "step": 6646 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.244, + "step": 6647 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 6648 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.2225, + "step": 6649 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.172, + "step": 6650 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1895, + "step": 6651 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.171, + "step": 6652 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1577, + "step": 6653 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.203, + "step": 6654 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 6655 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 6656 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1755, + "step": 6657 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.2286, + "step": 6658 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1888, + "step": 6659 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.2393, + "step": 6660 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.2591, + "step": 6661 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.2058, + "step": 6662 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1878, + "step": 6663 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.1672, + "step": 6664 + }, + { + "epoch": 0.4, + "learning_rate": 2e-05, + "loss": 0.2329, + "step": 6665 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.2442, + "step": 6666 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 6667 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 6668 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 6669 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1524, + "step": 6670 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.193, + "step": 6671 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 6672 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1742, + "step": 6673 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1846, + "step": 6674 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1938, + "step": 6675 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1935, + "step": 6676 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.2444, + "step": 6677 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1716, + "step": 6678 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.225, + "step": 6679 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.198, + "step": 6680 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1718, + "step": 6681 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1816, + "step": 6682 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 6683 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1641, + "step": 6684 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.2484, + "step": 6685 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1684, + "step": 6686 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.2126, + "step": 6687 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1626, + "step": 6688 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1832, + "step": 6689 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 6690 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1659, + "step": 6691 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 6692 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1666, + "step": 6693 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 6694 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.167, + "step": 6695 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1933, + "step": 6696 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 6697 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1608, + "step": 6698 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1679, + "step": 6699 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.3217, + "step": 6700 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1825, + "step": 6701 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1663, + "step": 6702 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1873, + "step": 6703 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 6704 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1748, + "step": 6705 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.2023, + "step": 6706 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.2041, + "step": 6707 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 6708 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.183, + "step": 6709 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 6710 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 6711 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.2019, + "step": 6712 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1911, + "step": 6713 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 6714 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1739, + "step": 6715 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.188, + "step": 6716 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 6717 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.2169, + "step": 6718 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1854, + "step": 6719 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1798, + "step": 6720 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1986, + "step": 6721 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1564, + "step": 6722 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1797, + "step": 6723 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 6724 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1801, + "step": 6725 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 6726 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1818, + "step": 6727 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 6728 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1899, + "step": 6729 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.2399, + "step": 6730 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1832, + "step": 6731 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1881, + "step": 6732 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1503, + "step": 6733 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 6734 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1462, + "step": 6735 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1885, + "step": 6736 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 6737 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1989, + "step": 6738 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 6739 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.202, + "step": 6740 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 6741 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1597, + "step": 6742 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 6743 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1947, + "step": 6744 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.2434, + "step": 6745 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.2146, + "step": 6746 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1702, + "step": 6747 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 6748 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 6749 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 6750 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1623, + "step": 6751 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.2012, + "step": 6752 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.2006, + "step": 6753 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1933, + "step": 6754 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1626, + "step": 6755 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 6756 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 6757 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 6758 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1485, + "step": 6759 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1768, + "step": 6760 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.2102, + "step": 6761 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1696, + "step": 6762 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.2204, + "step": 6763 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1855, + "step": 6764 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1647, + "step": 6765 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.2017, + "step": 6766 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 6767 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.213, + "step": 6768 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 6769 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1628, + "step": 6770 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1796, + "step": 6771 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 6772 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1796, + "step": 6773 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 6774 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.2116, + "step": 6775 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1578, + "step": 6776 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1889, + "step": 6777 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1742, + "step": 6778 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1882, + "step": 6779 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1852, + "step": 6780 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1873, + "step": 6781 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1771, + "step": 6782 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.152, + "step": 6783 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1824, + "step": 6784 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 6785 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.2038, + "step": 6786 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.208, + "step": 6787 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.2033, + "step": 6788 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.2146, + "step": 6789 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 6790 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1691, + "step": 6791 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1513, + "step": 6792 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1665, + "step": 6793 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.2371, + "step": 6794 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1644, + "step": 6795 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1936, + "step": 6796 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1862, + "step": 6797 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.175, + "step": 6798 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1665, + "step": 6799 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1837, + "step": 6800 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1635, + "step": 6801 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1719, + "step": 6802 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 6803 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1685, + "step": 6804 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1822, + "step": 6805 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1957, + "step": 6806 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1447, + "step": 6807 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1799, + "step": 6808 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1648, + "step": 6809 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1708, + "step": 6810 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 6811 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 6812 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1554, + "step": 6813 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.188, + "step": 6814 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1843, + "step": 6815 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 6816 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1687, + "step": 6817 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1801, + "step": 6818 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1711, + "step": 6819 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.168, + "step": 6820 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1752, + "step": 6821 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1757, + "step": 6822 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.2114, + "step": 6823 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1988, + "step": 6824 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.2132, + "step": 6825 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 6826 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1985, + "step": 6827 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1728, + "step": 6828 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1765, + "step": 6829 + }, + { + "epoch": 0.41, + "learning_rate": 2e-05, + "loss": 0.1835, + "step": 6830 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.2067, + "step": 6831 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1833, + "step": 6832 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.154, + "step": 6833 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1831, + "step": 6834 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1546, + "step": 6835 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 6836 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1803, + "step": 6837 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1711, + "step": 6838 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1785, + "step": 6839 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1845, + "step": 6840 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1977, + "step": 6841 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 6842 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1818, + "step": 6843 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.179, + "step": 6844 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1804, + "step": 6845 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 6846 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1915, + "step": 6847 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.206, + "step": 6848 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1709, + "step": 6849 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1789, + "step": 6850 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1759, + "step": 6851 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.2032, + "step": 6852 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1905, + "step": 6853 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1583, + "step": 6854 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1972, + "step": 6855 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1577, + "step": 6856 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 6857 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1657, + "step": 6858 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1762, + "step": 6859 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 6860 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1804, + "step": 6861 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1958, + "step": 6862 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.2768, + "step": 6863 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1599, + "step": 6864 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 6865 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.168, + "step": 6866 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1754, + "step": 6867 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1632, + "step": 6868 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 6869 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1746, + "step": 6870 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1721, + "step": 6871 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1796, + "step": 6872 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 6873 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1897, + "step": 6874 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1924, + "step": 6875 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 6876 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1882, + "step": 6877 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 6878 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1916, + "step": 6879 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 6880 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.192, + "step": 6881 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.221, + "step": 6882 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1764, + "step": 6883 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 6884 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1695, + "step": 6885 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1983, + "step": 6886 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 6887 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1851, + "step": 6888 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1743, + "step": 6889 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 6890 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1835, + "step": 6891 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.2266, + "step": 6892 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1824, + "step": 6893 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.2063, + "step": 6894 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1756, + "step": 6895 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 6896 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1965, + "step": 6897 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1873, + "step": 6898 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.2099, + "step": 6899 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1705, + "step": 6900 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1901, + "step": 6901 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 6902 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.227, + "step": 6903 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1961, + "step": 6904 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1745, + "step": 6905 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.2021, + "step": 6906 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.2, + "step": 6907 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.2251, + "step": 6908 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1517, + "step": 6909 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1683, + "step": 6910 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1968, + "step": 6911 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1916, + "step": 6912 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 6913 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.2028, + "step": 6914 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1532, + "step": 6915 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.2106, + "step": 6916 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1872, + "step": 6917 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1606, + "step": 6918 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1937, + "step": 6919 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1989, + "step": 6920 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.2214, + "step": 6921 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1651, + "step": 6922 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1723, + "step": 6923 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1993, + "step": 6924 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1885, + "step": 6925 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1669, + "step": 6926 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1692, + "step": 6927 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1919, + "step": 6928 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 6929 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1659, + "step": 6930 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1791, + "step": 6931 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.2181, + "step": 6932 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1525, + "step": 6933 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1734, + "step": 6934 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1734, + "step": 6935 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1667, + "step": 6936 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.2334, + "step": 6937 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 6938 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1713, + "step": 6939 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.2297, + "step": 6940 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1658, + "step": 6941 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1673, + "step": 6942 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1691, + "step": 6943 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.163, + "step": 6944 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1673, + "step": 6945 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1634, + "step": 6946 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1673, + "step": 6947 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1549, + "step": 6948 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1778, + "step": 6949 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1897, + "step": 6950 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 6951 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.2081, + "step": 6952 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1823, + "step": 6953 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 6954 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1695, + "step": 6955 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1661, + "step": 6956 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.2027, + "step": 6957 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1923, + "step": 6958 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 6959 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1554, + "step": 6960 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.17, + "step": 6961 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1655, + "step": 6962 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1845, + "step": 6963 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.196, + "step": 6964 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1633, + "step": 6965 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.186, + "step": 6966 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.2053, + "step": 6967 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1596, + "step": 6968 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 6969 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1775, + "step": 6970 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.167, + "step": 6971 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1838, + "step": 6972 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1913, + "step": 6973 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.2371, + "step": 6974 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.174, + "step": 6975 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.2268, + "step": 6976 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.2041, + "step": 6977 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1772, + "step": 6978 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1638, + "step": 6979 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.2188, + "step": 6980 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 6981 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1969, + "step": 6982 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1754, + "step": 6983 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.2013, + "step": 6984 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.247, + "step": 6985 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.174, + "step": 6986 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1519, + "step": 6987 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1709, + "step": 6988 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 6989 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1655, + "step": 6990 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 6991 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1766, + "step": 6992 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.1675, + "step": 6993 + }, + { + "epoch": 0.42, + "learning_rate": 2e-05, + "loss": 0.2592, + "step": 6994 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1975, + "step": 6995 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1549, + "step": 6996 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.193, + "step": 6997 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1862, + "step": 6998 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1576, + "step": 6999 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.165, + "step": 7000 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.236, + "step": 7001 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.2145, + "step": 7002 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1695, + "step": 7003 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 7004 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.2183, + "step": 7005 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 7006 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1746, + "step": 7007 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.167, + "step": 7008 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1794, + "step": 7009 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 7010 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1768, + "step": 7011 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1588, + "step": 7012 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.171, + "step": 7013 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1859, + "step": 7014 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1734, + "step": 7015 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1652, + "step": 7016 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 7017 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 7018 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 7019 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1971, + "step": 7020 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1779, + "step": 7021 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1722, + "step": 7022 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 7023 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 7024 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 7025 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 7026 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.2225, + "step": 7027 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.2151, + "step": 7028 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1839, + "step": 7029 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 7030 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 7031 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1709, + "step": 7032 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 7033 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1918, + "step": 7034 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1761, + "step": 7035 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1778, + "step": 7036 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 7037 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.2103, + "step": 7038 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.178, + "step": 7039 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.2013, + "step": 7040 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1845, + "step": 7041 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.2075, + "step": 7042 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.2294, + "step": 7043 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1759, + "step": 7044 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1868, + "step": 7045 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.2081, + "step": 7046 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1874, + "step": 7047 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1918, + "step": 7048 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1816, + "step": 7049 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 7050 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.248, + "step": 7051 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1645, + "step": 7052 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 7053 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 7054 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.183, + "step": 7055 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1589, + "step": 7056 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.164, + "step": 7057 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 7058 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1815, + "step": 7059 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1821, + "step": 7060 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1898, + "step": 7061 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 7062 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1767, + "step": 7063 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1878, + "step": 7064 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1818, + "step": 7065 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 7066 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 7067 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1511, + "step": 7068 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.25, + "step": 7069 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 7070 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1896, + "step": 7071 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1913, + "step": 7072 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1706, + "step": 7073 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 7074 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 7075 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 7076 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1931, + "step": 7077 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1813, + "step": 7078 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1883, + "step": 7079 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1756, + "step": 7080 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1589, + "step": 7081 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.2131, + "step": 7082 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1583, + "step": 7083 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1727, + "step": 7084 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1903, + "step": 7085 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 7086 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1991, + "step": 7087 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.2443, + "step": 7088 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 7089 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.196, + "step": 7090 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1765, + "step": 7091 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.2158, + "step": 7092 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1857, + "step": 7093 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 7094 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1883, + "step": 7095 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.2051, + "step": 7096 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1605, + "step": 7097 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1519, + "step": 7098 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.2362, + "step": 7099 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1972, + "step": 7100 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1657, + "step": 7101 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 7102 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.2195, + "step": 7103 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 7104 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 7105 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1848, + "step": 7106 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1974, + "step": 7107 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.221, + "step": 7108 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1992, + "step": 7109 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.2237, + "step": 7110 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1918, + "step": 7111 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.2284, + "step": 7112 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1595, + "step": 7113 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1794, + "step": 7114 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1623, + "step": 7115 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1813, + "step": 7116 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 7117 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.2299, + "step": 7118 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1754, + "step": 7119 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1602, + "step": 7120 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1733, + "step": 7121 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1667, + "step": 7122 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 7123 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 7124 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.2056, + "step": 7125 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1637, + "step": 7126 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.155, + "step": 7127 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1897, + "step": 7128 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1646, + "step": 7129 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1654, + "step": 7130 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1527, + "step": 7131 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1739, + "step": 7132 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1552, + "step": 7133 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.2282, + "step": 7134 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1948, + "step": 7135 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 7136 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1817, + "step": 7137 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1443, + "step": 7138 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1903, + "step": 7139 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.2093, + "step": 7140 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1561, + "step": 7141 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1987, + "step": 7142 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 7143 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1593, + "step": 7144 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 7145 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1646, + "step": 7146 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1907, + "step": 7147 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.168, + "step": 7148 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.2216, + "step": 7149 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1813, + "step": 7150 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.2196, + "step": 7151 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.2015, + "step": 7152 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.175, + "step": 7153 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.2055, + "step": 7154 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.2309, + "step": 7155 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.158, + "step": 7156 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1771, + "step": 7157 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.2096, + "step": 7158 + }, + { + "epoch": 0.43, + "learning_rate": 2e-05, + "loss": 0.1803, + "step": 7159 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1922, + "step": 7160 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1812, + "step": 7161 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1683, + "step": 7162 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1643, + "step": 7163 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1498, + "step": 7164 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1623, + "step": 7165 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.2136, + "step": 7166 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 7167 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1825, + "step": 7168 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1886, + "step": 7169 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.2089, + "step": 7170 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.172, + "step": 7171 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1722, + "step": 7172 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 7173 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1641, + "step": 7174 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1605, + "step": 7175 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.2231, + "step": 7176 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 7177 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.2207, + "step": 7178 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.2138, + "step": 7179 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1715, + "step": 7180 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 7181 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 7182 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1894, + "step": 7183 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1779, + "step": 7184 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1623, + "step": 7185 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1802, + "step": 7186 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 7187 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1784, + "step": 7188 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1625, + "step": 7189 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 7190 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1714, + "step": 7191 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1672, + "step": 7192 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 7193 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1548, + "step": 7194 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 7195 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1611, + "step": 7196 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 7197 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.2226, + "step": 7198 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1528, + "step": 7199 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.2214, + "step": 7200 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 7201 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1784, + "step": 7202 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 7203 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1734, + "step": 7204 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1916, + "step": 7205 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1683, + "step": 7206 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1724, + "step": 7207 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 7208 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.2318, + "step": 7209 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1956, + "step": 7210 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.164, + "step": 7211 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 7212 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1741, + "step": 7213 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.226, + "step": 7214 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1772, + "step": 7215 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1533, + "step": 7216 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1987, + "step": 7217 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1709, + "step": 7218 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1754, + "step": 7219 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 7220 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1743, + "step": 7221 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.2252, + "step": 7222 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.2026, + "step": 7223 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.2, + "step": 7224 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1734, + "step": 7225 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 7226 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1721, + "step": 7227 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.2102, + "step": 7228 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.161, + "step": 7229 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.2429, + "step": 7230 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1442, + "step": 7231 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1843, + "step": 7232 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 7233 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1617, + "step": 7234 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1939, + "step": 7235 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 7236 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1519, + "step": 7237 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1827, + "step": 7238 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 7239 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1767, + "step": 7240 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 7241 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1515, + "step": 7242 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1987, + "step": 7243 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 7244 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.2187, + "step": 7245 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 7246 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1743, + "step": 7247 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1752, + "step": 7248 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1451, + "step": 7249 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.2305, + "step": 7250 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 7251 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1966, + "step": 7252 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.2012, + "step": 7253 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 7254 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 7255 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1551, + "step": 7256 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1733, + "step": 7257 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1943, + "step": 7258 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.186, + "step": 7259 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.2175, + "step": 7260 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.237, + "step": 7261 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 7262 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1526, + "step": 7263 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1555, + "step": 7264 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 7265 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.156, + "step": 7266 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 7267 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1588, + "step": 7268 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 7269 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1621, + "step": 7270 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1925, + "step": 7271 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1998, + "step": 7272 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1815, + "step": 7273 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1563, + "step": 7274 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1869, + "step": 7275 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.2283, + "step": 7276 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.156, + "step": 7277 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 7278 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1945, + "step": 7279 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1763, + "step": 7280 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.17, + "step": 7281 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.2356, + "step": 7282 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1902, + "step": 7283 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.204, + "step": 7284 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.164, + "step": 7285 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1847, + "step": 7286 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1785, + "step": 7287 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 7288 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 7289 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.155, + "step": 7290 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 7291 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1991, + "step": 7292 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1476, + "step": 7293 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1814, + "step": 7294 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 7295 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 7296 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1822, + "step": 7297 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1722, + "step": 7298 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.247, + "step": 7299 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 7300 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1507, + "step": 7301 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 7302 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 7303 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1817, + "step": 7304 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1876, + "step": 7305 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.2078, + "step": 7306 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1741, + "step": 7307 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1675, + "step": 7308 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 7309 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1931, + "step": 7310 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1635, + "step": 7311 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1692, + "step": 7312 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 7313 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1989, + "step": 7314 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 7315 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.167, + "step": 7316 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1715, + "step": 7317 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1714, + "step": 7318 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.2461, + "step": 7319 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.2222, + "step": 7320 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.2088, + "step": 7321 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1658, + "step": 7322 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1607, + "step": 7323 + }, + { + "epoch": 0.44, + "learning_rate": 2e-05, + "loss": 0.1765, + "step": 7324 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1801, + "step": 7325 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1967, + "step": 7326 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1805, + "step": 7327 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.251, + "step": 7328 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1876, + "step": 7329 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.19, + "step": 7330 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1745, + "step": 7331 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1881, + "step": 7332 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1695, + "step": 7333 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1804, + "step": 7334 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1692, + "step": 7335 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1624, + "step": 7336 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1673, + "step": 7337 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1639, + "step": 7338 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1691, + "step": 7339 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1538, + "step": 7340 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1816, + "step": 7341 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1849, + "step": 7342 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 7343 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1562, + "step": 7344 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1634, + "step": 7345 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1934, + "step": 7346 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 7347 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.2015, + "step": 7348 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.2119, + "step": 7349 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1568, + "step": 7350 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1656, + "step": 7351 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1779, + "step": 7352 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1696, + "step": 7353 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 7354 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.2227, + "step": 7355 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1451, + "step": 7356 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1557, + "step": 7357 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.2331, + "step": 7358 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1786, + "step": 7359 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1687, + "step": 7360 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1751, + "step": 7361 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1626, + "step": 7362 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.2441, + "step": 7363 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 7364 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1704, + "step": 7365 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 7366 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.2188, + "step": 7367 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1894, + "step": 7368 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.17, + "step": 7369 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.2136, + "step": 7370 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1703, + "step": 7371 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1931, + "step": 7372 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1716, + "step": 7373 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.2101, + "step": 7374 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1855, + "step": 7375 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1534, + "step": 7376 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1998, + "step": 7377 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 7378 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1983, + "step": 7379 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1558, + "step": 7380 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.2059, + "step": 7381 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 7382 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 7383 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.2048, + "step": 7384 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.2479, + "step": 7385 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 7386 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1538, + "step": 7387 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 7388 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1791, + "step": 7389 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1962, + "step": 7390 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1718, + "step": 7391 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.217, + "step": 7392 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 7393 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 7394 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1716, + "step": 7395 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.2022, + "step": 7396 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.2064, + "step": 7397 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 7398 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.2014, + "step": 7399 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1666, + "step": 7400 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1682, + "step": 7401 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.2208, + "step": 7402 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 7403 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1649, + "step": 7404 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1552, + "step": 7405 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 7406 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.2159, + "step": 7407 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1621, + "step": 7408 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1553, + "step": 7409 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1661, + "step": 7410 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.2007, + "step": 7411 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1696, + "step": 7412 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1856, + "step": 7413 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1817, + "step": 7414 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.2032, + "step": 7415 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.2078, + "step": 7416 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1889, + "step": 7417 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1881, + "step": 7418 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1714, + "step": 7419 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1777, + "step": 7420 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 7421 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1695, + "step": 7422 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 7423 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1971, + "step": 7424 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1513, + "step": 7425 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 7426 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1639, + "step": 7427 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1541, + "step": 7428 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1952, + "step": 7429 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 7430 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1805, + "step": 7431 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1719, + "step": 7432 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.2118, + "step": 7433 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1885, + "step": 7434 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1944, + "step": 7435 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.2565, + "step": 7436 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1597, + "step": 7437 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 7438 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1698, + "step": 7439 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1895, + "step": 7440 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1702, + "step": 7441 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1681, + "step": 7442 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.2065, + "step": 7443 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1689, + "step": 7444 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.205, + "step": 7445 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.213, + "step": 7446 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1926, + "step": 7447 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.2003, + "step": 7448 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1782, + "step": 7449 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.2033, + "step": 7450 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 7451 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1678, + "step": 7452 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1753, + "step": 7453 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1708, + "step": 7454 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 7455 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 7456 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1876, + "step": 7457 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.2241, + "step": 7458 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1447, + "step": 7459 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.158, + "step": 7460 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1549, + "step": 7461 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1682, + "step": 7462 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 7463 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.154, + "step": 7464 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 7465 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1717, + "step": 7466 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1916, + "step": 7467 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1943, + "step": 7468 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 7469 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1906, + "step": 7470 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1844, + "step": 7471 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1519, + "step": 7472 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 7473 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1535, + "step": 7474 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.2087, + "step": 7475 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1524, + "step": 7476 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1669, + "step": 7477 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1916, + "step": 7478 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.2237, + "step": 7479 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1876, + "step": 7480 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1538, + "step": 7481 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1964, + "step": 7482 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1672, + "step": 7483 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1774, + "step": 7484 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.189, + "step": 7485 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1678, + "step": 7486 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1743, + "step": 7487 + }, + { + "epoch": 0.45, + "learning_rate": 2e-05, + "loss": 0.1985, + "step": 7488 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1681, + "step": 7489 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1598, + "step": 7490 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1769, + "step": 7491 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1917, + "step": 7492 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.2153, + "step": 7493 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1672, + "step": 7494 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1732, + "step": 7495 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.2445, + "step": 7496 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 7497 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.2009, + "step": 7498 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.2026, + "step": 7499 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1736, + "step": 7500 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1709, + "step": 7501 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1701, + "step": 7502 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 7503 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1807, + "step": 7504 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1708, + "step": 7505 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1548, + "step": 7506 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.2224, + "step": 7507 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1812, + "step": 7508 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 7509 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 7510 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1853, + "step": 7511 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1934, + "step": 7512 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1659, + "step": 7513 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 7514 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 7515 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.186, + "step": 7516 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1777, + "step": 7517 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.179, + "step": 7518 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.2157, + "step": 7519 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1498, + "step": 7520 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1538, + "step": 7521 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1603, + "step": 7522 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 7523 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 7524 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1762, + "step": 7525 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1647, + "step": 7526 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1927, + "step": 7527 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 7528 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.2057, + "step": 7529 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 7530 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1671, + "step": 7531 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1987, + "step": 7532 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 7533 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1693, + "step": 7534 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1734, + "step": 7535 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 7536 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1675, + "step": 7537 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.2013, + "step": 7538 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1721, + "step": 7539 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1611, + "step": 7540 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 7541 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1648, + "step": 7542 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1989, + "step": 7543 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1633, + "step": 7544 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1948, + "step": 7545 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.152, + "step": 7546 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1604, + "step": 7547 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1887, + "step": 7548 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1981, + "step": 7549 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.2177, + "step": 7550 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1766, + "step": 7551 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.193, + "step": 7552 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1894, + "step": 7553 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 7554 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.2067, + "step": 7555 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 7556 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 7557 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.261, + "step": 7558 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1997, + "step": 7559 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1889, + "step": 7560 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1745, + "step": 7561 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 7562 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1842, + "step": 7563 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1851, + "step": 7564 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1617, + "step": 7565 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1986, + "step": 7566 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1611, + "step": 7567 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1734, + "step": 7568 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 7569 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1823, + "step": 7570 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 7571 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1456, + "step": 7572 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 7573 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1879, + "step": 7574 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 7575 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1682, + "step": 7576 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1738, + "step": 7577 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1676, + "step": 7578 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.2051, + "step": 7579 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 7580 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.2065, + "step": 7581 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1796, + "step": 7582 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1832, + "step": 7583 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1828, + "step": 7584 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 7585 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1914, + "step": 7586 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 7587 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1667, + "step": 7588 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1823, + "step": 7589 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1711, + "step": 7590 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1795, + "step": 7591 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1662, + "step": 7592 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1853, + "step": 7593 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.2086, + "step": 7594 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1748, + "step": 7595 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1616, + "step": 7596 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1738, + "step": 7597 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 7598 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1862, + "step": 7599 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1744, + "step": 7600 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.2164, + "step": 7601 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1659, + "step": 7602 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1834, + "step": 7603 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.2283, + "step": 7604 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1592, + "step": 7605 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 7606 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 7607 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.177, + "step": 7608 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 7609 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1711, + "step": 7610 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1906, + "step": 7611 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1937, + "step": 7612 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.2004, + "step": 7613 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1933, + "step": 7614 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1871, + "step": 7615 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.203, + "step": 7616 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.238, + "step": 7617 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 7618 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.2273, + "step": 7619 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 7620 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1814, + "step": 7621 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.2009, + "step": 7622 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.167, + "step": 7623 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.163, + "step": 7624 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1696, + "step": 7625 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1582, + "step": 7626 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1649, + "step": 7627 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 7628 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.2364, + "step": 7629 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1593, + "step": 7630 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 7631 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 7632 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1786, + "step": 7633 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1555, + "step": 7634 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1894, + "step": 7635 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 7636 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1803, + "step": 7637 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1714, + "step": 7638 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.152, + "step": 7639 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.2133, + "step": 7640 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1442, + "step": 7641 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 7642 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 7643 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.2023, + "step": 7644 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 7645 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1764, + "step": 7646 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 7647 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.2084, + "step": 7648 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1871, + "step": 7649 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.2381, + "step": 7650 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 7651 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1525, + "step": 7652 + }, + { + "epoch": 0.46, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 7653 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.2342, + "step": 7654 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1839, + "step": 7655 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.184, + "step": 7656 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 7657 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1847, + "step": 7658 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1676, + "step": 7659 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 7660 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1934, + "step": 7661 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1775, + "step": 7662 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1736, + "step": 7663 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1791, + "step": 7664 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.2093, + "step": 7665 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1579, + "step": 7666 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1552, + "step": 7667 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1717, + "step": 7668 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1718, + "step": 7669 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.177, + "step": 7670 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1553, + "step": 7671 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1928, + "step": 7672 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.179, + "step": 7673 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 7674 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1553, + "step": 7675 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1777, + "step": 7676 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1585, + "step": 7677 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 7678 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1669, + "step": 7679 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.2023, + "step": 7680 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1902, + "step": 7681 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.165, + "step": 7682 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 7683 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1865, + "step": 7684 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1553, + "step": 7685 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.165, + "step": 7686 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1845, + "step": 7687 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1833, + "step": 7688 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1765, + "step": 7689 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.2084, + "step": 7690 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1698, + "step": 7691 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1858, + "step": 7692 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1944, + "step": 7693 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1752, + "step": 7694 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1645, + "step": 7695 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1963, + "step": 7696 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 7697 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1667, + "step": 7698 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1613, + "step": 7699 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.2239, + "step": 7700 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1785, + "step": 7701 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1663, + "step": 7702 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1904, + "step": 7703 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1524, + "step": 7704 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.2012, + "step": 7705 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1853, + "step": 7706 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1945, + "step": 7707 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.2058, + "step": 7708 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.2066, + "step": 7709 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1841, + "step": 7710 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1799, + "step": 7711 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 7712 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 7713 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.2003, + "step": 7714 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 7715 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1831, + "step": 7716 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 7717 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 7718 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1628, + "step": 7719 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1635, + "step": 7720 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1779, + "step": 7721 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1722, + "step": 7722 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 7723 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 7724 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1746, + "step": 7725 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.2206, + "step": 7726 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.167, + "step": 7727 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.2028, + "step": 7728 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.2057, + "step": 7729 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 7730 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1595, + "step": 7731 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.145, + "step": 7732 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1577, + "step": 7733 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 7734 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.183, + "step": 7735 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1919, + "step": 7736 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1984, + "step": 7737 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.2027, + "step": 7738 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1854, + "step": 7739 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1589, + "step": 7740 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1788, + "step": 7741 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.2085, + "step": 7742 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1497, + "step": 7743 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 7744 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1705, + "step": 7745 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 7746 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.188, + "step": 7747 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1932, + "step": 7748 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1911, + "step": 7749 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 7750 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 7751 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1857, + "step": 7752 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.2352, + "step": 7753 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1549, + "step": 7754 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.2016, + "step": 7755 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1717, + "step": 7756 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.2153, + "step": 7757 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1937, + "step": 7758 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1953, + "step": 7759 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.164, + "step": 7760 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 7761 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1576, + "step": 7762 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.221, + "step": 7763 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 7764 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.2025, + "step": 7765 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1904, + "step": 7766 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1629, + "step": 7767 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1783, + "step": 7768 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1672, + "step": 7769 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1817, + "step": 7770 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1666, + "step": 7771 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1642, + "step": 7772 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1995, + "step": 7773 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.2342, + "step": 7774 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1743, + "step": 7775 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.189, + "step": 7776 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 7777 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1646, + "step": 7778 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 7779 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1722, + "step": 7780 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.2058, + "step": 7781 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1557, + "step": 7782 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1722, + "step": 7783 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1647, + "step": 7784 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.2063, + "step": 7785 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1975, + "step": 7786 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1817, + "step": 7787 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1869, + "step": 7788 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1497, + "step": 7789 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.2029, + "step": 7790 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1751, + "step": 7791 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1876, + "step": 7792 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1687, + "step": 7793 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.2025, + "step": 7794 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1625, + "step": 7795 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1895, + "step": 7796 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1943, + "step": 7797 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1913, + "step": 7798 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 7799 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.161, + "step": 7800 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1792, + "step": 7801 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 7802 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1824, + "step": 7803 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1827, + "step": 7804 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1852, + "step": 7805 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 7806 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 7807 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.2096, + "step": 7808 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.2075, + "step": 7809 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1595, + "step": 7810 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 7811 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 7812 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1614, + "step": 7813 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.145, + "step": 7814 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 7815 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 7816 + }, + { + "epoch": 0.47, + "learning_rate": 2e-05, + "loss": 0.1864, + "step": 7817 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 7818 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 7819 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1465, + "step": 7820 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1683, + "step": 7821 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1866, + "step": 7822 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1909, + "step": 7823 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.155, + "step": 7824 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.2258, + "step": 7825 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1723, + "step": 7826 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.163, + "step": 7827 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1567, + "step": 7828 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1628, + "step": 7829 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1868, + "step": 7830 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1666, + "step": 7831 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1821, + "step": 7832 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1829, + "step": 7833 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1757, + "step": 7834 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1598, + "step": 7835 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.178, + "step": 7836 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 7837 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1947, + "step": 7838 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.2034, + "step": 7839 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 7840 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 7841 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.159, + "step": 7842 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1958, + "step": 7843 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.2108, + "step": 7844 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1886, + "step": 7845 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1683, + "step": 7846 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1993, + "step": 7847 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1639, + "step": 7848 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1669, + "step": 7849 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1668, + "step": 7850 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.2255, + "step": 7851 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.225, + "step": 7852 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1666, + "step": 7853 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 7854 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1646, + "step": 7855 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1653, + "step": 7856 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1786, + "step": 7857 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1732, + "step": 7858 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1488, + "step": 7859 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.2191, + "step": 7860 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.2271, + "step": 7861 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1951, + "step": 7862 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1688, + "step": 7863 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 7864 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1916, + "step": 7865 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1623, + "step": 7866 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1745, + "step": 7867 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1775, + "step": 7868 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1751, + "step": 7869 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 7870 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 7871 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1819, + "step": 7872 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 7873 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1592, + "step": 7874 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1945, + "step": 7875 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1558, + "step": 7876 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.2081, + "step": 7877 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1913, + "step": 7878 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.186, + "step": 7879 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1909, + "step": 7880 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1875, + "step": 7881 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1725, + "step": 7882 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 7883 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1848, + "step": 7884 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1495, + "step": 7885 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.2036, + "step": 7886 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 7887 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.16, + "step": 7888 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1774, + "step": 7889 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 7890 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1967, + "step": 7891 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 7892 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.178, + "step": 7893 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.2359, + "step": 7894 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1978, + "step": 7895 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1515, + "step": 7896 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1801, + "step": 7897 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1646, + "step": 7898 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1696, + "step": 7899 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1596, + "step": 7900 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1781, + "step": 7901 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.2051, + "step": 7902 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.158, + "step": 7903 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1924, + "step": 7904 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.2579, + "step": 7905 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1854, + "step": 7906 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1857, + "step": 7907 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1686, + "step": 7908 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1872, + "step": 7909 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.2398, + "step": 7910 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 7911 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1675, + "step": 7912 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1851, + "step": 7913 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.147, + "step": 7914 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.2031, + "step": 7915 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.2105, + "step": 7916 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1902, + "step": 7917 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1926, + "step": 7918 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.205, + "step": 7919 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 7920 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1888, + "step": 7921 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1625, + "step": 7922 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1932, + "step": 7923 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1526, + "step": 7924 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1784, + "step": 7925 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1557, + "step": 7926 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 7927 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1914, + "step": 7928 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1725, + "step": 7929 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 7930 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1626, + "step": 7931 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1787, + "step": 7932 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 7933 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1887, + "step": 7934 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1614, + "step": 7935 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 7936 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 7937 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1938, + "step": 7938 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1988, + "step": 7939 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1695, + "step": 7940 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1803, + "step": 7941 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1906, + "step": 7942 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1674, + "step": 7943 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.2074, + "step": 7944 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1635, + "step": 7945 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 7946 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1851, + "step": 7947 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 7948 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1851, + "step": 7949 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 7950 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 7951 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 7952 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1582, + "step": 7953 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.2045, + "step": 7954 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1598, + "step": 7955 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1797, + "step": 7956 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1902, + "step": 7957 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 7958 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1988, + "step": 7959 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 7960 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1623, + "step": 7961 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.2307, + "step": 7962 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.2078, + "step": 7963 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1656, + "step": 7964 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1852, + "step": 7965 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1936, + "step": 7966 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1456, + "step": 7967 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1895, + "step": 7968 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 7969 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1826, + "step": 7970 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1548, + "step": 7971 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.177, + "step": 7972 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1822, + "step": 7973 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.2265, + "step": 7974 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1989, + "step": 7975 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.2206, + "step": 7976 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1796, + "step": 7977 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 7978 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1802, + "step": 7979 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.2483, + "step": 7980 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1909, + "step": 7981 + }, + { + "epoch": 0.48, + "learning_rate": 2e-05, + "loss": 0.1928, + "step": 7982 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1892, + "step": 7983 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1794, + "step": 7984 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1948, + "step": 7985 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1796, + "step": 7986 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1805, + "step": 7987 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1684, + "step": 7988 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1673, + "step": 7989 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.2582, + "step": 7990 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 7991 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 7992 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1796, + "step": 7993 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 7994 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1652, + "step": 7995 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 7996 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1818, + "step": 7997 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 7998 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1596, + "step": 7999 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 8000 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1698, + "step": 8001 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1909, + "step": 8002 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 8003 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1668, + "step": 8004 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 8005 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1627, + "step": 8006 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1909, + "step": 8007 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 8008 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.2049, + "step": 8009 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1676, + "step": 8010 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1658, + "step": 8011 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 8012 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 8013 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1943, + "step": 8014 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.2088, + "step": 8015 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1645, + "step": 8016 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 8017 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1554, + "step": 8018 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.192, + "step": 8019 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.2203, + "step": 8020 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1574, + "step": 8021 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1872, + "step": 8022 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1914, + "step": 8023 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1876, + "step": 8024 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1864, + "step": 8025 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.2147, + "step": 8026 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1651, + "step": 8027 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1672, + "step": 8028 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.2032, + "step": 8029 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1663, + "step": 8030 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.158, + "step": 8031 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1735, + "step": 8032 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.2154, + "step": 8033 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 8034 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 8035 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.2107, + "step": 8036 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1741, + "step": 8037 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1696, + "step": 8038 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 8039 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.2071, + "step": 8040 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 8041 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.2169, + "step": 8042 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1901, + "step": 8043 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.2008, + "step": 8044 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1753, + "step": 8045 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1519, + "step": 8046 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1456, + "step": 8047 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.167, + "step": 8048 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1777, + "step": 8049 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.2308, + "step": 8050 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1648, + "step": 8051 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1965, + "step": 8052 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 8053 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1822, + "step": 8054 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1927, + "step": 8055 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1995, + "step": 8056 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 8057 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.2117, + "step": 8058 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1541, + "step": 8059 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 8060 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 8061 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.2405, + "step": 8062 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1645, + "step": 8063 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1576, + "step": 8064 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1819, + "step": 8065 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 8066 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1919, + "step": 8067 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1591, + "step": 8068 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.2128, + "step": 8069 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1558, + "step": 8070 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1995, + "step": 8071 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.2087, + "step": 8072 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.2095, + "step": 8073 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.2505, + "step": 8074 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1951, + "step": 8075 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1733, + "step": 8076 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1665, + "step": 8077 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1697, + "step": 8078 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1851, + "step": 8079 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 8080 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1459, + "step": 8081 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.2079, + "step": 8082 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1888, + "step": 8083 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 8084 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 8085 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.2018, + "step": 8086 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.2012, + "step": 8087 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1777, + "step": 8088 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1551, + "step": 8089 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.2105, + "step": 8090 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1651, + "step": 8091 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.2087, + "step": 8092 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 8093 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1857, + "step": 8094 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1784, + "step": 8095 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 8096 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1599, + "step": 8097 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1922, + "step": 8098 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1924, + "step": 8099 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.18, + "step": 8100 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1915, + "step": 8101 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1736, + "step": 8102 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.2188, + "step": 8103 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1756, + "step": 8104 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1686, + "step": 8105 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.2215, + "step": 8106 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 8107 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 8108 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1758, + "step": 8109 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1761, + "step": 8110 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1549, + "step": 8111 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1686, + "step": 8112 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1722, + "step": 8113 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1451, + "step": 8114 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1836, + "step": 8115 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1725, + "step": 8116 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1674, + "step": 8117 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1558, + "step": 8118 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.2124, + "step": 8119 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1808, + "step": 8120 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 8121 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1778, + "step": 8122 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 8123 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1983, + "step": 8124 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1468, + "step": 8125 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1485, + "step": 8126 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1884, + "step": 8127 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1799, + "step": 8128 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 8129 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.2129, + "step": 8130 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1519, + "step": 8131 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 8132 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 8133 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1976, + "step": 8134 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1532, + "step": 8135 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1725, + "step": 8136 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1517, + "step": 8137 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.2003, + "step": 8138 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1784, + "step": 8139 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 8140 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1468, + "step": 8141 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1745, + "step": 8142 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1555, + "step": 8143 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1732, + "step": 8144 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1578, + "step": 8145 + }, + { + "epoch": 0.49, + "learning_rate": 2e-05, + "loss": 0.1716, + "step": 8146 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1799, + "step": 8147 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 8148 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 8149 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1946, + "step": 8150 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1618, + "step": 8151 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1502, + "step": 8152 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1633, + "step": 8153 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1733, + "step": 8154 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.2021, + "step": 8155 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 8156 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1855, + "step": 8157 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1638, + "step": 8158 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 8159 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.228, + "step": 8160 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1848, + "step": 8161 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 8162 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.181, + "step": 8163 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 8164 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1855, + "step": 8165 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1502, + "step": 8166 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.165, + "step": 8167 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 8168 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1699, + "step": 8169 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1577, + "step": 8170 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 8171 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1897, + "step": 8172 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1999, + "step": 8173 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1853, + "step": 8174 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 8175 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1757, + "step": 8176 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1923, + "step": 8177 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1955, + "step": 8178 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1968, + "step": 8179 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.2014, + "step": 8180 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1832, + "step": 8181 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 8182 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.2069, + "step": 8183 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 8184 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.2098, + "step": 8185 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1585, + "step": 8186 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1683, + "step": 8187 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1677, + "step": 8188 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.194, + "step": 8189 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.184, + "step": 8190 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.18, + "step": 8191 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1798, + "step": 8192 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.2168, + "step": 8193 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1744, + "step": 8194 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1761, + "step": 8195 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1578, + "step": 8196 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 8197 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1694, + "step": 8198 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 8199 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1614, + "step": 8200 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.2152, + "step": 8201 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1498, + "step": 8202 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 8203 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1996, + "step": 8204 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1679, + "step": 8205 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1838, + "step": 8206 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1782, + "step": 8207 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.207, + "step": 8208 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1829, + "step": 8209 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1854, + "step": 8210 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1842, + "step": 8211 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.2224, + "step": 8212 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1718, + "step": 8213 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1776, + "step": 8214 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.2231, + "step": 8215 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1746, + "step": 8216 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1971, + "step": 8217 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.2013, + "step": 8218 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1639, + "step": 8219 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1895, + "step": 8220 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 8221 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 8222 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.2056, + "step": 8223 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1908, + "step": 8224 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 8225 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.2296, + "step": 8226 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1851, + "step": 8227 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1549, + "step": 8228 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 8229 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1713, + "step": 8230 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1762, + "step": 8231 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 8232 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1823, + "step": 8233 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 8234 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1808, + "step": 8235 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1706, + "step": 8236 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1861, + "step": 8237 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1956, + "step": 8238 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1607, + "step": 8239 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1965, + "step": 8240 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1582, + "step": 8241 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1843, + "step": 8242 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1785, + "step": 8243 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.2142, + "step": 8244 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1527, + "step": 8245 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1694, + "step": 8246 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1725, + "step": 8247 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.182, + "step": 8248 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1747, + "step": 8249 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1671, + "step": 8250 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1459, + "step": 8251 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1927, + "step": 8252 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 8253 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 8254 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1682, + "step": 8255 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1515, + "step": 8256 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 8257 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1542, + "step": 8258 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.2025, + "step": 8259 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1761, + "step": 8260 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1848, + "step": 8261 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1639, + "step": 8262 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.2125, + "step": 8263 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.2052, + "step": 8264 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1909, + "step": 8265 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1761, + "step": 8266 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 8267 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1695, + "step": 8268 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.171, + "step": 8269 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1703, + "step": 8270 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1823, + "step": 8271 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 8272 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 8273 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1588, + "step": 8274 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 8275 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1797, + "step": 8276 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1577, + "step": 8277 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.184, + "step": 8278 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1649, + "step": 8279 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1954, + "step": 8280 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1782, + "step": 8281 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 8282 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1687, + "step": 8283 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1564, + "step": 8284 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 8285 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1941, + "step": 8286 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1704, + "step": 8287 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1661, + "step": 8288 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1958, + "step": 8289 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1726, + "step": 8290 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 8291 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.2412, + "step": 8292 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1782, + "step": 8293 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 8294 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 8295 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1853, + "step": 8296 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 8297 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 8298 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.2039, + "step": 8299 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.2087, + "step": 8300 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1759, + "step": 8301 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1938, + "step": 8302 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1669, + "step": 8303 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.2266, + "step": 8304 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1874, + "step": 8305 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1997, + "step": 8306 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1617, + "step": 8307 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 8308 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.2118, + "step": 8309 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 8310 + }, + { + "epoch": 0.5, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 8311 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1691, + "step": 8312 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 8313 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1875, + "step": 8314 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1841, + "step": 8315 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1785, + "step": 8316 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1944, + "step": 8317 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1961, + "step": 8318 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1925, + "step": 8319 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.197, + "step": 8320 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1614, + "step": 8321 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.2085, + "step": 8322 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1666, + "step": 8323 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1918, + "step": 8324 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 8325 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1603, + "step": 8326 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.2148, + "step": 8327 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1655, + "step": 8328 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 8329 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.145, + "step": 8330 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 8331 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.19, + "step": 8332 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1866, + "step": 8333 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1718, + "step": 8334 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.163, + "step": 8335 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.2178, + "step": 8336 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1936, + "step": 8337 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1503, + "step": 8338 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.2072, + "step": 8339 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 8340 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1756, + "step": 8341 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.2084, + "step": 8342 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.21, + "step": 8343 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.183, + "step": 8344 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.213, + "step": 8345 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.164, + "step": 8346 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1771, + "step": 8347 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.2168, + "step": 8348 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1939, + "step": 8349 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1867, + "step": 8350 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1607, + "step": 8351 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1527, + "step": 8352 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.2028, + "step": 8353 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.2046, + "step": 8354 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1672, + "step": 8355 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1631, + "step": 8356 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 8357 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 8358 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1845, + "step": 8359 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1592, + "step": 8360 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.2103, + "step": 8361 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 8362 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.2311, + "step": 8363 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.2096, + "step": 8364 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 8365 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.192, + "step": 8366 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1671, + "step": 8367 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 8368 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1846, + "step": 8369 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1787, + "step": 8370 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.2079, + "step": 8371 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1744, + "step": 8372 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1945, + "step": 8373 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.2441, + "step": 8374 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1878, + "step": 8375 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1727, + "step": 8376 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1913, + "step": 8377 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 8378 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1631, + "step": 8379 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 8380 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 8381 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 8382 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 8383 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 8384 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1952, + "step": 8385 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 8386 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1617, + "step": 8387 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1674, + "step": 8388 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1562, + "step": 8389 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 8390 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 8391 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 8392 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.2039, + "step": 8393 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.154, + "step": 8394 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1578, + "step": 8395 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 8396 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1704, + "step": 8397 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1942, + "step": 8398 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1698, + "step": 8399 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1796, + "step": 8400 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1809, + "step": 8401 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 8402 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1746, + "step": 8403 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 8404 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1649, + "step": 8405 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1648, + "step": 8406 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 8407 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.2135, + "step": 8408 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.192, + "step": 8409 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1694, + "step": 8410 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 8411 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 8412 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 8413 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.2658, + "step": 8414 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 8415 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.2219, + "step": 8416 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.152, + "step": 8417 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 8418 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.189, + "step": 8419 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1875, + "step": 8420 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1952, + "step": 8421 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1791, + "step": 8422 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 8423 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.169, + "step": 8424 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1723, + "step": 8425 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 8426 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 8427 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1849, + "step": 8428 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1746, + "step": 8429 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1706, + "step": 8430 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.154, + "step": 8431 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1842, + "step": 8432 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.2284, + "step": 8433 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1962, + "step": 8434 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 8435 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1754, + "step": 8436 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1554, + "step": 8437 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.163, + "step": 8438 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1564, + "step": 8439 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.178, + "step": 8440 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1741, + "step": 8441 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1987, + "step": 8442 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1815, + "step": 8443 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 8444 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.2122, + "step": 8445 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 8446 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 8447 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1758, + "step": 8448 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1684, + "step": 8449 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1577, + "step": 8450 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1679, + "step": 8451 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1739, + "step": 8452 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1551, + "step": 8453 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1675, + "step": 8454 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 8455 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1728, + "step": 8456 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.2017, + "step": 8457 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1921, + "step": 8458 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1839, + "step": 8459 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 8460 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 8461 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.236, + "step": 8462 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 8463 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1856, + "step": 8464 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.2047, + "step": 8465 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1659, + "step": 8466 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.2273, + "step": 8467 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1672, + "step": 8468 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1623, + "step": 8469 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1867, + "step": 8470 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1447, + "step": 8471 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1578, + "step": 8472 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1704, + "step": 8473 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.2218, + "step": 8474 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1664, + "step": 8475 + }, + { + "epoch": 0.51, + "learning_rate": 2e-05, + "loss": 0.1778, + "step": 8476 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 8477 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1903, + "step": 8478 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.2143, + "step": 8479 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.2022, + "step": 8480 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1704, + "step": 8481 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1895, + "step": 8482 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1669, + "step": 8483 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.2035, + "step": 8484 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.2146, + "step": 8485 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1959, + "step": 8486 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 8487 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.2101, + "step": 8488 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1874, + "step": 8489 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.2089, + "step": 8490 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1682, + "step": 8491 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 8492 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1965, + "step": 8493 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1626, + "step": 8494 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1541, + "step": 8495 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1631, + "step": 8496 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1833, + "step": 8497 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1658, + "step": 8498 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1749, + "step": 8499 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 8500 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1798, + "step": 8501 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1902, + "step": 8502 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 8503 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1926, + "step": 8504 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 8505 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1624, + "step": 8506 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.2281, + "step": 8507 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1643, + "step": 8508 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1672, + "step": 8509 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1857, + "step": 8510 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.2115, + "step": 8511 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.2052, + "step": 8512 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.19, + "step": 8513 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1604, + "step": 8514 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1691, + "step": 8515 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 8516 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 8517 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1776, + "step": 8518 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1576, + "step": 8519 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 8520 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1852, + "step": 8521 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1903, + "step": 8522 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1818, + "step": 8523 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 8524 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1914, + "step": 8525 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1894, + "step": 8526 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.19, + "step": 8527 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1708, + "step": 8528 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 8529 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1652, + "step": 8530 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.176, + "step": 8531 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1698, + "step": 8532 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1956, + "step": 8533 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1818, + "step": 8534 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.2075, + "step": 8535 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1527, + "step": 8536 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1672, + "step": 8537 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1909, + "step": 8538 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1561, + "step": 8539 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1507, + "step": 8540 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1878, + "step": 8541 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.2035, + "step": 8542 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1499, + "step": 8543 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1568, + "step": 8544 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1573, + "step": 8545 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1804, + "step": 8546 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.2049, + "step": 8547 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1777, + "step": 8548 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 8549 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 8550 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 8551 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.163, + "step": 8552 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1905, + "step": 8553 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 8554 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.2505, + "step": 8555 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1623, + "step": 8556 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1697, + "step": 8557 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1913, + "step": 8558 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1591, + "step": 8559 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 8560 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.2068, + "step": 8561 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1793, + "step": 8562 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1792, + "step": 8563 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1656, + "step": 8564 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 8565 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.2066, + "step": 8566 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1857, + "step": 8567 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1719, + "step": 8568 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1583, + "step": 8569 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1601, + "step": 8570 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1821, + "step": 8571 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1764, + "step": 8572 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1717, + "step": 8573 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.19, + "step": 8574 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 8575 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1639, + "step": 8576 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.2496, + "step": 8577 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1647, + "step": 8578 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1524, + "step": 8579 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.2012, + "step": 8580 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1675, + "step": 8581 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 8582 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.2097, + "step": 8583 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.17, + "step": 8584 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1629, + "step": 8585 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.2012, + "step": 8586 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.2063, + "step": 8587 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1714, + "step": 8588 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 8589 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 8590 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 8591 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1583, + "step": 8592 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1827, + "step": 8593 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1848, + "step": 8594 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 8595 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.2025, + "step": 8596 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1671, + "step": 8597 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1447, + "step": 8598 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1722, + "step": 8599 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1858, + "step": 8600 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1766, + "step": 8601 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1601, + "step": 8602 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1792, + "step": 8603 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.2029, + "step": 8604 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1644, + "step": 8605 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1618, + "step": 8606 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.2139, + "step": 8607 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1658, + "step": 8608 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 8609 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.2094, + "step": 8610 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.172, + "step": 8611 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1669, + "step": 8612 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 8613 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1825, + "step": 8614 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1582, + "step": 8615 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.2314, + "step": 8616 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 8617 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1755, + "step": 8618 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1691, + "step": 8619 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1794, + "step": 8620 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.148, + "step": 8621 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1687, + "step": 8622 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 8623 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 8624 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1668, + "step": 8625 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 8626 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.2177, + "step": 8627 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1946, + "step": 8628 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1797, + "step": 8629 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1629, + "step": 8630 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.2173, + "step": 8631 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 8632 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1718, + "step": 8633 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1465, + "step": 8634 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 8635 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1667, + "step": 8636 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1562, + "step": 8637 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1789, + "step": 8638 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1908, + "step": 8639 + }, + { + "epoch": 0.52, + "learning_rate": 2e-05, + "loss": 0.1821, + "step": 8640 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.2017, + "step": 8641 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.2076, + "step": 8642 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.187, + "step": 8643 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1907, + "step": 8644 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 8645 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1626, + "step": 8646 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.2072, + "step": 8647 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.2064, + "step": 8648 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1749, + "step": 8649 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 8650 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1706, + "step": 8651 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 8652 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.182, + "step": 8653 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 8654 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1708, + "step": 8655 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 8656 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 8657 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 8658 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1652, + "step": 8659 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1527, + "step": 8660 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.186, + "step": 8661 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 8662 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 8663 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1993, + "step": 8664 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1944, + "step": 8665 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 8666 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.158, + "step": 8667 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1751, + "step": 8668 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1878, + "step": 8669 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1804, + "step": 8670 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 8671 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1765, + "step": 8672 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 8673 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.2555, + "step": 8674 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1732, + "step": 8675 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.17, + "step": 8676 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1598, + "step": 8677 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.162, + "step": 8678 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1665, + "step": 8679 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.2013, + "step": 8680 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1971, + "step": 8681 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.2086, + "step": 8682 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1555, + "step": 8683 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1524, + "step": 8684 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1574, + "step": 8685 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1734, + "step": 8686 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.2064, + "step": 8687 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.2205, + "step": 8688 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 8689 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1965, + "step": 8690 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 8691 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.2062, + "step": 8692 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 8693 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.2269, + "step": 8694 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 8695 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1861, + "step": 8696 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 8697 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.2077, + "step": 8698 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1787, + "step": 8699 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.172, + "step": 8700 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.148, + "step": 8701 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 8702 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1796, + "step": 8703 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 8704 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 8705 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1599, + "step": 8706 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 8707 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 8708 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1925, + "step": 8709 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1803, + "step": 8710 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1502, + "step": 8711 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1654, + "step": 8712 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1862, + "step": 8713 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1824, + "step": 8714 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 8715 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1785, + "step": 8716 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.2218, + "step": 8717 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.154, + "step": 8718 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.164, + "step": 8719 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1682, + "step": 8720 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1887, + "step": 8721 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1829, + "step": 8722 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1888, + "step": 8723 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1628, + "step": 8724 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1528, + "step": 8725 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.2109, + "step": 8726 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1688, + "step": 8727 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1542, + "step": 8728 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1875, + "step": 8729 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1882, + "step": 8730 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1691, + "step": 8731 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1907, + "step": 8732 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1905, + "step": 8733 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1829, + "step": 8734 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1807, + "step": 8735 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.2213, + "step": 8736 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1669, + "step": 8737 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.198, + "step": 8738 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1686, + "step": 8739 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1694, + "step": 8740 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1686, + "step": 8741 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1831, + "step": 8742 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1971, + "step": 8743 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 8744 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 8745 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1826, + "step": 8746 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1549, + "step": 8747 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1977, + "step": 8748 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1969, + "step": 8749 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 8750 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.2008, + "step": 8751 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1655, + "step": 8752 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.2188, + "step": 8753 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1538, + "step": 8754 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 8755 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1592, + "step": 8756 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 8757 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.178, + "step": 8758 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1857, + "step": 8759 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 8760 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1476, + "step": 8761 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1548, + "step": 8762 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 8763 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 8764 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 8765 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1889, + "step": 8766 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1818, + "step": 8767 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.2058, + "step": 8768 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1655, + "step": 8769 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.2109, + "step": 8770 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1552, + "step": 8771 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1653, + "step": 8772 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1669, + "step": 8773 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1562, + "step": 8774 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1796, + "step": 8775 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 8776 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1846, + "step": 8777 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.198, + "step": 8778 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1611, + "step": 8779 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1809, + "step": 8780 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1787, + "step": 8781 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 8782 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.159, + "step": 8783 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 8784 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1939, + "step": 8785 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1562, + "step": 8786 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 8787 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1447, + "step": 8788 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 8789 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.2248, + "step": 8790 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.2065, + "step": 8791 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1785, + "step": 8792 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1736, + "step": 8793 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1469, + "step": 8794 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1557, + "step": 8795 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 8796 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 8797 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1597, + "step": 8798 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 8799 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 8800 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.176, + "step": 8801 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 8802 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1765, + "step": 8803 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.1497, + "step": 8804 + }, + { + "epoch": 0.53, + "learning_rate": 2e-05, + "loss": 0.2006, + "step": 8805 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 8806 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 8807 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.189, + "step": 8808 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 8809 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1833, + "step": 8810 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1755, + "step": 8811 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1603, + "step": 8812 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.158, + "step": 8813 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1937, + "step": 8814 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1785, + "step": 8815 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1673, + "step": 8816 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 8817 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1674, + "step": 8818 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1708, + "step": 8819 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 8820 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1707, + "step": 8821 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 8822 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1866, + "step": 8823 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.2143, + "step": 8824 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.2045, + "step": 8825 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1705, + "step": 8826 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1782, + "step": 8827 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 8828 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1505, + "step": 8829 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.2022, + "step": 8830 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1722, + "step": 8831 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 8832 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 8833 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 8834 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1614, + "step": 8835 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1912, + "step": 8836 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.2235, + "step": 8837 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1908, + "step": 8838 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1699, + "step": 8839 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.2115, + "step": 8840 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1574, + "step": 8841 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 8842 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 8843 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.2035, + "step": 8844 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1574, + "step": 8845 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1781, + "step": 8846 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1831, + "step": 8847 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1715, + "step": 8848 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1873, + "step": 8849 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 8850 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.159, + "step": 8851 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 8852 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 8853 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1785, + "step": 8854 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.2009, + "step": 8855 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.2094, + "step": 8856 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1856, + "step": 8857 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1857, + "step": 8858 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1732, + "step": 8859 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.2084, + "step": 8860 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.2021, + "step": 8861 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 8862 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.2058, + "step": 8863 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 8864 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.2156, + "step": 8865 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 8866 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 8867 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.2072, + "step": 8868 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1821, + "step": 8869 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 8870 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 8871 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1686, + "step": 8872 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1859, + "step": 8873 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 8874 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1684, + "step": 8875 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1954, + "step": 8876 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 8877 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1887, + "step": 8878 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1516, + "step": 8879 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1673, + "step": 8880 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1603, + "step": 8881 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.2207, + "step": 8882 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 8883 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.194, + "step": 8884 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 8885 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 8886 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 8887 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 8888 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1597, + "step": 8889 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1667, + "step": 8890 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 8891 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 8892 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.2047, + "step": 8893 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1893, + "step": 8894 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1626, + "step": 8895 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.196, + "step": 8896 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 8897 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 8898 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 8899 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.2335, + "step": 8900 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1716, + "step": 8901 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1633, + "step": 8902 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1591, + "step": 8903 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 8904 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 8905 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1786, + "step": 8906 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 8907 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1774, + "step": 8908 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.171, + "step": 8909 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.2036, + "step": 8910 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 8911 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1709, + "step": 8912 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1771, + "step": 8913 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1456, + "step": 8914 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 8915 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 8916 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1743, + "step": 8917 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1733, + "step": 8918 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1836, + "step": 8919 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1606, + "step": 8920 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1741, + "step": 8921 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.2071, + "step": 8922 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1635, + "step": 8923 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1601, + "step": 8924 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1652, + "step": 8925 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1579, + "step": 8926 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1749, + "step": 8927 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.162, + "step": 8928 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1502, + "step": 8929 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 8930 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1942, + "step": 8931 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1718, + "step": 8932 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 8933 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 8934 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1618, + "step": 8935 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 8936 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.2047, + "step": 8937 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1776, + "step": 8938 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1808, + "step": 8939 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1926, + "step": 8940 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1901, + "step": 8941 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 8942 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1578, + "step": 8943 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1861, + "step": 8944 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1635, + "step": 8945 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.2041, + "step": 8946 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.209, + "step": 8947 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1744, + "step": 8948 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.154, + "step": 8949 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1515, + "step": 8950 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 8951 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1889, + "step": 8952 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.159, + "step": 8953 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 8954 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1926, + "step": 8955 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.218, + "step": 8956 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1934, + "step": 8957 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1629, + "step": 8958 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 8959 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1925, + "step": 8960 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1614, + "step": 8961 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.2001, + "step": 8962 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1752, + "step": 8963 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1735, + "step": 8964 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1521, + "step": 8965 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1655, + "step": 8966 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 8967 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 8968 + }, + { + "epoch": 0.54, + "learning_rate": 2e-05, + "loss": 0.1786, + "step": 8969 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.2236, + "step": 8970 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1555, + "step": 8971 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.188, + "step": 8972 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1785, + "step": 8973 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.194, + "step": 8974 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 8975 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 8976 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 8977 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1841, + "step": 8978 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 8979 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1968, + "step": 8980 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 8981 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1601, + "step": 8982 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 8983 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 8984 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 8985 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1766, + "step": 8986 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.2096, + "step": 8987 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 8988 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1622, + "step": 8989 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1519, + "step": 8990 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1815, + "step": 8991 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 8992 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1864, + "step": 8993 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1812, + "step": 8994 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 8995 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1633, + "step": 8996 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1733, + "step": 8997 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 8998 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1525, + "step": 8999 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 9000 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1732, + "step": 9001 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1775, + "step": 9002 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 9003 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 9004 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.2063, + "step": 9005 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 9006 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1498, + "step": 9007 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 9008 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 9009 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 9010 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.2099, + "step": 9011 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.186, + "step": 9012 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1823, + "step": 9013 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.18, + "step": 9014 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1854, + "step": 9015 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 9016 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1593, + "step": 9017 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.2, + "step": 9018 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 9019 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.2063, + "step": 9020 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.2083, + "step": 9021 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 9022 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1753, + "step": 9023 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1991, + "step": 9024 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.175, + "step": 9025 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1686, + "step": 9026 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1729, + "step": 9027 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1465, + "step": 9028 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1562, + "step": 9029 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1772, + "step": 9030 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.177, + "step": 9031 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1817, + "step": 9032 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1694, + "step": 9033 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.2008, + "step": 9034 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1759, + "step": 9035 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1677, + "step": 9036 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1671, + "step": 9037 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1456, + "step": 9038 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1711, + "step": 9039 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1534, + "step": 9040 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1871, + "step": 9041 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 9042 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.2004, + "step": 9043 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1747, + "step": 9044 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1903, + "step": 9045 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1819, + "step": 9046 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1774, + "step": 9047 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1563, + "step": 9048 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1519, + "step": 9049 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 9050 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1541, + "step": 9051 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.182, + "step": 9052 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 9053 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1665, + "step": 9054 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 9055 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1611, + "step": 9056 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 9057 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1724, + "step": 9058 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.2209, + "step": 9059 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1627, + "step": 9060 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1707, + "step": 9061 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 9062 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.171, + "step": 9063 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.154, + "step": 9064 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1789, + "step": 9065 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1694, + "step": 9066 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1622, + "step": 9067 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.16, + "step": 9068 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 9069 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1771, + "step": 9070 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1608, + "step": 9071 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 9072 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.2288, + "step": 9073 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1578, + "step": 9074 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 9075 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1459, + "step": 9076 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.2109, + "step": 9077 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 9078 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1884, + "step": 9079 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1874, + "step": 9080 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 9081 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.2172, + "step": 9082 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.2345, + "step": 9083 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.147, + "step": 9084 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.171, + "step": 9085 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1873, + "step": 9086 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1642, + "step": 9087 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.2221, + "step": 9088 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1923, + "step": 9089 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 9090 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1643, + "step": 9091 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1583, + "step": 9092 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.182, + "step": 9093 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 9094 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1798, + "step": 9095 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1813, + "step": 9096 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1712, + "step": 9097 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 9098 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1869, + "step": 9099 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1577, + "step": 9100 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.2077, + "step": 9101 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.2219, + "step": 9102 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1626, + "step": 9103 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1997, + "step": 9104 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 9105 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 9106 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1588, + "step": 9107 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1814, + "step": 9108 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 9109 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.2375, + "step": 9110 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 9111 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1813, + "step": 9112 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1843, + "step": 9113 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 9114 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1634, + "step": 9115 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1929, + "step": 9116 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.2181, + "step": 9117 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 9118 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 9119 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1998, + "step": 9120 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1517, + "step": 9121 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1675, + "step": 9122 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 9123 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1862, + "step": 9124 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.2077, + "step": 9125 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1858, + "step": 9126 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 9127 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1576, + "step": 9128 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1573, + "step": 9129 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1928, + "step": 9130 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1975, + "step": 9131 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.2043, + "step": 9132 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 9133 + }, + { + "epoch": 0.55, + "learning_rate": 2e-05, + "loss": 0.1603, + "step": 9134 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 9135 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 9136 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 9137 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1856, + "step": 9138 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1673, + "step": 9139 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1591, + "step": 9140 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 9141 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 9142 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.2087, + "step": 9143 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 9144 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.2104, + "step": 9145 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1651, + "step": 9146 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.2044, + "step": 9147 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1598, + "step": 9148 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1793, + "step": 9149 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 9150 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.171, + "step": 9151 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1863, + "step": 9152 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.2218, + "step": 9153 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1656, + "step": 9154 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1808, + "step": 9155 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1663, + "step": 9156 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1685, + "step": 9157 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1773, + "step": 9158 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.2134, + "step": 9159 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1644, + "step": 9160 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1811, + "step": 9161 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1647, + "step": 9162 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 9163 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 9164 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1822, + "step": 9165 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.2011, + "step": 9166 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1672, + "step": 9167 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 9168 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 9169 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 9170 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1499, + "step": 9171 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.2178, + "step": 9172 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 9173 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1829, + "step": 9174 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 9175 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1896, + "step": 9176 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 9177 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 9178 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1653, + "step": 9179 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1624, + "step": 9180 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1747, + "step": 9181 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1811, + "step": 9182 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1948, + "step": 9183 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1911, + "step": 9184 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1549, + "step": 9185 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1926, + "step": 9186 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1625, + "step": 9187 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1906, + "step": 9188 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 9189 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.2212, + "step": 9190 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 9191 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 9192 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1613, + "step": 9193 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1911, + "step": 9194 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1737, + "step": 9195 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 9196 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 9197 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1857, + "step": 9198 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.17, + "step": 9199 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1538, + "step": 9200 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.2219, + "step": 9201 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1899, + "step": 9202 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1814, + "step": 9203 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.204, + "step": 9204 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 9205 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1807, + "step": 9206 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1739, + "step": 9207 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 9208 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 9209 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 9210 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.2005, + "step": 9211 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1809, + "step": 9212 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1478, + "step": 9213 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1978, + "step": 9214 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 9215 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1705, + "step": 9216 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1681, + "step": 9217 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 9218 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1901, + "step": 9219 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 9220 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1742, + "step": 9221 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 9222 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1933, + "step": 9223 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.153, + "step": 9224 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.2304, + "step": 9225 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1675, + "step": 9226 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1478, + "step": 9227 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1904, + "step": 9228 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1779, + "step": 9229 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.204, + "step": 9230 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1975, + "step": 9231 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1737, + "step": 9232 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.2032, + "step": 9233 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1456, + "step": 9234 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 9235 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 9236 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.2095, + "step": 9237 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 9238 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1709, + "step": 9239 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.2126, + "step": 9240 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.2289, + "step": 9241 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1792, + "step": 9242 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1969, + "step": 9243 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.187, + "step": 9244 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.181, + "step": 9245 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1727, + "step": 9246 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 9247 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1724, + "step": 9248 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1618, + "step": 9249 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1852, + "step": 9250 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1948, + "step": 9251 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.2008, + "step": 9252 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 9253 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 9254 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1676, + "step": 9255 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.197, + "step": 9256 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 9257 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 9258 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 9259 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 9260 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1892, + "step": 9261 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1769, + "step": 9262 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.188, + "step": 9263 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 9264 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 9265 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.2181, + "step": 9266 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1814, + "step": 9267 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 9268 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 9269 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 9270 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1756, + "step": 9271 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 9272 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1715, + "step": 9273 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1712, + "step": 9274 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 9275 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 9276 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1733, + "step": 9277 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1568, + "step": 9278 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1534, + "step": 9279 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1725, + "step": 9280 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1899, + "step": 9281 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.184, + "step": 9282 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 9283 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 9284 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 9285 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.2069, + "step": 9286 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1715, + "step": 9287 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1528, + "step": 9288 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1738, + "step": 9289 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1835, + "step": 9290 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1825, + "step": 9291 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.146, + "step": 9292 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1839, + "step": 9293 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.2053, + "step": 9294 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.2249, + "step": 9295 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1459, + "step": 9296 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1826, + "step": 9297 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1507, + "step": 9298 + }, + { + "epoch": 0.56, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 9299 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 9300 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 9301 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1968, + "step": 9302 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 9303 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1518, + "step": 9304 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1538, + "step": 9305 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1918, + "step": 9306 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1623, + "step": 9307 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1963, + "step": 9308 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.2288, + "step": 9309 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 9310 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 9311 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1585, + "step": 9312 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.2118, + "step": 9313 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 9314 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1542, + "step": 9315 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1631, + "step": 9316 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 9317 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 9318 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 9319 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 9320 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 9321 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 9322 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.159, + "step": 9323 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1593, + "step": 9324 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1465, + "step": 9325 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1974, + "step": 9326 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1626, + "step": 9327 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 9328 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.2044, + "step": 9329 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.209, + "step": 9330 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.2239, + "step": 9331 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 9332 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1945, + "step": 9333 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1703, + "step": 9334 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1729, + "step": 9335 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 9336 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1582, + "step": 9337 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1897, + "step": 9338 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.144, + "step": 9339 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1583, + "step": 9340 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1771, + "step": 9341 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1582, + "step": 9342 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1859, + "step": 9343 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.183, + "step": 9344 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1989, + "step": 9345 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 9346 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1567, + "step": 9347 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1542, + "step": 9348 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.162, + "step": 9349 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1558, + "step": 9350 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.184, + "step": 9351 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1649, + "step": 9352 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1562, + "step": 9353 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 9354 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1955, + "step": 9355 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1607, + "step": 9356 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.2009, + "step": 9357 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1815, + "step": 9358 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1845, + "step": 9359 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1563, + "step": 9360 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 9361 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 9362 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 9363 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.2045, + "step": 9364 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1561, + "step": 9365 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.2049, + "step": 9366 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1897, + "step": 9367 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1833, + "step": 9368 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1731, + "step": 9369 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.181, + "step": 9370 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1917, + "step": 9371 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1949, + "step": 9372 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.175, + "step": 9373 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1911, + "step": 9374 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1576, + "step": 9375 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 9376 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1513, + "step": 9377 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1695, + "step": 9378 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1714, + "step": 9379 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 9380 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1579, + "step": 9381 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 9382 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1826, + "step": 9383 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1713, + "step": 9384 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 9385 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 9386 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1627, + "step": 9387 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1597, + "step": 9388 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1701, + "step": 9389 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1798, + "step": 9390 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1684, + "step": 9391 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1555, + "step": 9392 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 9393 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1468, + "step": 9394 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 9395 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1576, + "step": 9396 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1468, + "step": 9397 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.172, + "step": 9398 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1595, + "step": 9399 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 9400 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1765, + "step": 9401 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1588, + "step": 9402 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 9403 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1845, + "step": 9404 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.2132, + "step": 9405 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1644, + "step": 9406 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1937, + "step": 9407 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1626, + "step": 9408 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.2124, + "step": 9409 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1459, + "step": 9410 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 9411 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 9412 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 9413 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1959, + "step": 9414 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1922, + "step": 9415 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1819, + "step": 9416 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1667, + "step": 9417 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.2131, + "step": 9418 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 9419 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1459, + "step": 9420 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.212, + "step": 9421 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1577, + "step": 9422 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1948, + "step": 9423 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1697, + "step": 9424 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1568, + "step": 9425 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.2126, + "step": 9426 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1724, + "step": 9427 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1563, + "step": 9428 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.2121, + "step": 9429 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 9430 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 9431 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.161, + "step": 9432 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1497, + "step": 9433 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 9434 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 9435 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.207, + "step": 9436 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 9437 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1519, + "step": 9438 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.2078, + "step": 9439 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.2013, + "step": 9440 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.162, + "step": 9441 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1764, + "step": 9442 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.2084, + "step": 9443 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 9444 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1993, + "step": 9445 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1567, + "step": 9446 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.188, + "step": 9447 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1828, + "step": 9448 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.2016, + "step": 9449 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1692, + "step": 9450 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.147, + "step": 9451 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 9452 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1849, + "step": 9453 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1793, + "step": 9454 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1623, + "step": 9455 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.2091, + "step": 9456 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 9457 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1605, + "step": 9458 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1628, + "step": 9459 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 9460 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1772, + "step": 9461 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1824, + "step": 9462 + }, + { + "epoch": 0.57, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 9463 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 9464 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1742, + "step": 9465 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.2025, + "step": 9466 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.184, + "step": 9467 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1616, + "step": 9468 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1679, + "step": 9469 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 9470 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1524, + "step": 9471 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 9472 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 9473 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1741, + "step": 9474 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1861, + "step": 9475 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1632, + "step": 9476 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1708, + "step": 9477 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1857, + "step": 9478 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 9479 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1831, + "step": 9480 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1854, + "step": 9481 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 9482 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 9483 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 9484 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.2001, + "step": 9485 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1753, + "step": 9486 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1897, + "step": 9487 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.235, + "step": 9488 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1708, + "step": 9489 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1823, + "step": 9490 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 9491 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1857, + "step": 9492 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1806, + "step": 9493 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1869, + "step": 9494 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1766, + "step": 9495 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 9496 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1876, + "step": 9497 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1758, + "step": 9498 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1511, + "step": 9499 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1774, + "step": 9500 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1642, + "step": 9501 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1713, + "step": 9502 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1891, + "step": 9503 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.161, + "step": 9504 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.2175, + "step": 9505 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1764, + "step": 9506 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1621, + "step": 9507 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 9508 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1641, + "step": 9509 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1867, + "step": 9510 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1662, + "step": 9511 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.2055, + "step": 9512 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1681, + "step": 9513 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 9514 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.177, + "step": 9515 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 9516 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 9517 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.2069, + "step": 9518 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1779, + "step": 9519 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 9520 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1832, + "step": 9521 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1748, + "step": 9522 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1742, + "step": 9523 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1611, + "step": 9524 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1729, + "step": 9525 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 9526 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1687, + "step": 9527 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.2065, + "step": 9528 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.188, + "step": 9529 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1552, + "step": 9530 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1785, + "step": 9531 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.2037, + "step": 9532 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 9533 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.193, + "step": 9534 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.2115, + "step": 9535 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.165, + "step": 9536 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 9537 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.203, + "step": 9538 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1524, + "step": 9539 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1816, + "step": 9540 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1777, + "step": 9541 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.2098, + "step": 9542 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.2013, + "step": 9543 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 9544 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 9545 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1833, + "step": 9546 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.174, + "step": 9547 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1675, + "step": 9548 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1641, + "step": 9549 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1747, + "step": 9550 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1775, + "step": 9551 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1958, + "step": 9552 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1709, + "step": 9553 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.175, + "step": 9554 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 9555 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.147, + "step": 9556 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 9557 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1763, + "step": 9558 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1809, + "step": 9559 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1927, + "step": 9560 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 9561 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 9562 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 9563 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1593, + "step": 9564 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 9565 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1905, + "step": 9566 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 9567 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1724, + "step": 9568 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1715, + "step": 9569 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1953, + "step": 9570 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 9571 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.2339, + "step": 9572 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.158, + "step": 9573 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1941, + "step": 9574 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1567, + "step": 9575 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 9576 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1825, + "step": 9577 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1811, + "step": 9578 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1668, + "step": 9579 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1916, + "step": 9580 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 9581 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1728, + "step": 9582 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.195, + "step": 9583 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1829, + "step": 9584 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1776, + "step": 9585 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1516, + "step": 9586 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1758, + "step": 9587 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.167, + "step": 9588 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1766, + "step": 9589 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1839, + "step": 9590 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.2033, + "step": 9591 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1931, + "step": 9592 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1669, + "step": 9593 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.2132, + "step": 9594 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1903, + "step": 9595 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1978, + "step": 9596 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 9597 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1997, + "step": 9598 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 9599 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1616, + "step": 9600 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.159, + "step": 9601 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.2174, + "step": 9602 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1429, + "step": 9603 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.167, + "step": 9604 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 9605 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1892, + "step": 9606 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1717, + "step": 9607 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1771, + "step": 9608 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 9609 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 9610 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1998, + "step": 9611 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.187, + "step": 9612 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 9613 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 9614 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 9615 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1947, + "step": 9616 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1754, + "step": 9617 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1629, + "step": 9618 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1974, + "step": 9619 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1736, + "step": 9620 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1666, + "step": 9621 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1554, + "step": 9622 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 9623 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1795, + "step": 9624 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 9625 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1656, + "step": 9626 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1666, + "step": 9627 + }, + { + "epoch": 0.58, + "learning_rate": 2e-05, + "loss": 0.1886, + "step": 9628 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.2194, + "step": 9629 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1677, + "step": 9630 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.2055, + "step": 9631 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1562, + "step": 9632 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1867, + "step": 9633 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1661, + "step": 9634 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1503, + "step": 9635 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1645, + "step": 9636 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1577, + "step": 9637 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1758, + "step": 9638 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1905, + "step": 9639 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.2085, + "step": 9640 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1701, + "step": 9641 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1626, + "step": 9642 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 9643 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1564, + "step": 9644 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1748, + "step": 9645 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.169, + "step": 9646 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1897, + "step": 9647 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1788, + "step": 9648 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1608, + "step": 9649 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 9650 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1721, + "step": 9651 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1681, + "step": 9652 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 9653 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 9654 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1637, + "step": 9655 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1732, + "step": 9656 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 9657 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 9658 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 9659 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1629, + "step": 9660 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1754, + "step": 9661 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1822, + "step": 9662 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.2461, + "step": 9663 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 9664 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1752, + "step": 9665 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1744, + "step": 9666 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1634, + "step": 9667 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1828, + "step": 9668 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1527, + "step": 9669 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1797, + "step": 9670 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.165, + "step": 9671 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1739, + "step": 9672 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1688, + "step": 9673 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.2203, + "step": 9674 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1521, + "step": 9675 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1701, + "step": 9676 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 9677 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1842, + "step": 9678 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.2189, + "step": 9679 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.227, + "step": 9680 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1561, + "step": 9681 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 9682 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 9683 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 9684 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.2049, + "step": 9685 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.159, + "step": 9686 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 9687 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1756, + "step": 9688 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 9689 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1911, + "step": 9690 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.2236, + "step": 9691 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1555, + "step": 9692 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1711, + "step": 9693 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1604, + "step": 9694 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1824, + "step": 9695 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1819, + "step": 9696 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1648, + "step": 9697 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1645, + "step": 9698 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1816, + "step": 9699 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1739, + "step": 9700 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1758, + "step": 9701 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1648, + "step": 9702 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1903, + "step": 9703 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1663, + "step": 9704 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1706, + "step": 9705 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 9706 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 9707 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1704, + "step": 9708 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.163, + "step": 9709 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1895, + "step": 9710 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1688, + "step": 9711 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1974, + "step": 9712 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1981, + "step": 9713 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1691, + "step": 9714 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1634, + "step": 9715 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1862, + "step": 9716 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.186, + "step": 9717 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.2384, + "step": 9718 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1553, + "step": 9719 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1868, + "step": 9720 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.2176, + "step": 9721 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1836, + "step": 9722 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.182, + "step": 9723 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1865, + "step": 9724 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1799, + "step": 9725 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 9726 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1642, + "step": 9727 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1842, + "step": 9728 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1595, + "step": 9729 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1662, + "step": 9730 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.191, + "step": 9731 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 9732 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.2343, + "step": 9733 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 9734 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1551, + "step": 9735 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1707, + "step": 9736 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1586, + "step": 9737 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 9738 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1848, + "step": 9739 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1542, + "step": 9740 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.2296, + "step": 9741 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 9742 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1805, + "step": 9743 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 9744 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1705, + "step": 9745 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.2026, + "step": 9746 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1824, + "step": 9747 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.2002, + "step": 9748 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1627, + "step": 9749 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 9750 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1927, + "step": 9751 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 9752 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 9753 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 9754 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1849, + "step": 9755 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1788, + "step": 9756 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 9757 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1605, + "step": 9758 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.221, + "step": 9759 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1669, + "step": 9760 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1694, + "step": 9761 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1733, + "step": 9762 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.206, + "step": 9763 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1552, + "step": 9764 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1605, + "step": 9765 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.156, + "step": 9766 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1877, + "step": 9767 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 9768 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 9769 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.16, + "step": 9770 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 9771 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1754, + "step": 9772 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.244, + "step": 9773 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1719, + "step": 9774 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1759, + "step": 9775 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 9776 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1702, + "step": 9777 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.2083, + "step": 9778 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 9779 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1748, + "step": 9780 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.2004, + "step": 9781 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1734, + "step": 9782 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1577, + "step": 9783 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.175, + "step": 9784 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1662, + "step": 9785 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 9786 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 9787 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 9788 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1914, + "step": 9789 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1809, + "step": 9790 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 9791 + }, + { + "epoch": 0.59, + "learning_rate": 2e-05, + "loss": 0.1746, + "step": 9792 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1808, + "step": 9793 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1803, + "step": 9794 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1953, + "step": 9795 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.2082, + "step": 9796 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1705, + "step": 9797 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1778, + "step": 9798 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 9799 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1784, + "step": 9800 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1892, + "step": 9801 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1549, + "step": 9802 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.208, + "step": 9803 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 9804 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.2227, + "step": 9805 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1621, + "step": 9806 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1798, + "step": 9807 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 9808 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1718, + "step": 9809 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 9810 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.2174, + "step": 9811 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1618, + "step": 9812 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1875, + "step": 9813 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.2233, + "step": 9814 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 9815 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1648, + "step": 9816 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 9817 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1663, + "step": 9818 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1789, + "step": 9819 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1875, + "step": 9820 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1677, + "step": 9821 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 9822 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 9823 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1873, + "step": 9824 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 9825 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1822, + "step": 9826 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1519, + "step": 9827 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 9828 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1535, + "step": 9829 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 9830 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1465, + "step": 9831 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1614, + "step": 9832 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 9833 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1716, + "step": 9834 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1507, + "step": 9835 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1793, + "step": 9836 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1554, + "step": 9837 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 9838 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 9839 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1935, + "step": 9840 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.2152, + "step": 9841 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 9842 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 9843 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.2014, + "step": 9844 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1759, + "step": 9845 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1878, + "step": 9846 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1948, + "step": 9847 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1694, + "step": 9848 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.2133, + "step": 9849 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 9850 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 9851 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1682, + "step": 9852 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1955, + "step": 9853 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1653, + "step": 9854 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1534, + "step": 9855 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1702, + "step": 9856 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 9857 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1869, + "step": 9858 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.168, + "step": 9859 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.2176, + "step": 9860 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1701, + "step": 9861 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.2411, + "step": 9862 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1815, + "step": 9863 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1503, + "step": 9864 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1579, + "step": 9865 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1974, + "step": 9866 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1923, + "step": 9867 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1511, + "step": 9868 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.2043, + "step": 9869 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1742, + "step": 9870 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1914, + "step": 9871 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.2142, + "step": 9872 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.2076, + "step": 9873 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.2134, + "step": 9874 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1469, + "step": 9875 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.195, + "step": 9876 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1922, + "step": 9877 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 9878 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 9879 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 9880 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.274, + "step": 9881 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1831, + "step": 9882 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.165, + "step": 9883 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1682, + "step": 9884 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 9885 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 9886 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1707, + "step": 9887 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1658, + "step": 9888 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1462, + "step": 9889 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1705, + "step": 9890 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1597, + "step": 9891 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 9892 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 9893 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1799, + "step": 9894 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1776, + "step": 9895 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 9896 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1733, + "step": 9897 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1909, + "step": 9898 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 9899 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.187, + "step": 9900 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1901, + "step": 9901 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1882, + "step": 9902 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1612, + "step": 9903 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.156, + "step": 9904 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1566, + "step": 9905 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 9906 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.163, + "step": 9907 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 9908 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.2093, + "step": 9909 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.146, + "step": 9910 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1747, + "step": 9911 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1697, + "step": 9912 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1595, + "step": 9913 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1505, + "step": 9914 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.2259, + "step": 9915 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.199, + "step": 9916 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1876, + "step": 9917 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 9918 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 9919 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1574, + "step": 9920 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.19, + "step": 9921 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 9922 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1674, + "step": 9923 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1956, + "step": 9924 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 9925 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1767, + "step": 9926 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1678, + "step": 9927 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1696, + "step": 9928 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.195, + "step": 9929 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 9930 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 9931 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 9932 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1724, + "step": 9933 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.2144, + "step": 9934 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1521, + "step": 9935 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1738, + "step": 9936 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1629, + "step": 9937 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1698, + "step": 9938 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.2028, + "step": 9939 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1682, + "step": 9940 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1762, + "step": 9941 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1718, + "step": 9942 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.177, + "step": 9943 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 9944 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1616, + "step": 9945 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.2296, + "step": 9946 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1714, + "step": 9947 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.2079, + "step": 9948 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.156, + "step": 9949 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1732, + "step": 9950 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 9951 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1994, + "step": 9952 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1753, + "step": 9953 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 9954 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 9955 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.1564, + "step": 9956 + }, + { + "epoch": 0.6, + "learning_rate": 2e-05, + "loss": 0.213, + "step": 9957 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1688, + "step": 9958 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1744, + "step": 9959 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 9960 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1588, + "step": 9961 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1623, + "step": 9962 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1723, + "step": 9963 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1763, + "step": 9964 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 9965 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1786, + "step": 9966 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.161, + "step": 9967 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1651, + "step": 9968 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1622, + "step": 9969 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1605, + "step": 9970 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1725, + "step": 9971 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.18, + "step": 9972 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 9973 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 9974 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1516, + "step": 9975 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.164, + "step": 9976 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1821, + "step": 9977 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 9978 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1631, + "step": 9979 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1821, + "step": 9980 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1741, + "step": 9981 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1685, + "step": 9982 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 9983 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1992, + "step": 9984 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.2042, + "step": 9985 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 9986 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.187, + "step": 9987 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1743, + "step": 9988 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1551, + "step": 9989 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.19, + "step": 9990 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 9991 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 9992 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 9993 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.2017, + "step": 9994 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.189, + "step": 9995 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 9996 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.2443, + "step": 9997 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.2168, + "step": 9998 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1983, + "step": 9999 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1789, + "step": 10000 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1606, + "step": 10001 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1666, + "step": 10002 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1887, + "step": 10003 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.2316, + "step": 10004 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 10005 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 10006 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.192, + "step": 10007 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.191, + "step": 10008 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1588, + "step": 10009 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1549, + "step": 10010 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 10011 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 10012 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1826, + "step": 10013 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.179, + "step": 10014 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1671, + "step": 10015 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.188, + "step": 10016 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1867, + "step": 10017 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1557, + "step": 10018 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1582, + "step": 10019 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1476, + "step": 10020 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.199, + "step": 10021 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.159, + "step": 10022 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1767, + "step": 10023 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1725, + "step": 10024 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 10025 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 10026 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 10027 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1704, + "step": 10028 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1542, + "step": 10029 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1746, + "step": 10030 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1642, + "step": 10031 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1503, + "step": 10032 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 10033 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 10034 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1817, + "step": 10035 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.2184, + "step": 10036 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1606, + "step": 10037 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1485, + "step": 10038 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1666, + "step": 10039 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 10040 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.175, + "step": 10041 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1794, + "step": 10042 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 10043 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.2293, + "step": 10044 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 10045 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 10046 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1884, + "step": 10047 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.17, + "step": 10048 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1662, + "step": 10049 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 10050 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1914, + "step": 10051 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 10052 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.2281, + "step": 10053 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1655, + "step": 10054 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1936, + "step": 10055 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 10056 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1682, + "step": 10057 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1638, + "step": 10058 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1998, + "step": 10059 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1502, + "step": 10060 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 10061 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 10062 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 10063 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 10064 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1622, + "step": 10065 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 10066 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 10067 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 10068 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 10069 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1787, + "step": 10070 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1839, + "step": 10071 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 10072 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 10073 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1932, + "step": 10074 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1617, + "step": 10075 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1567, + "step": 10076 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1851, + "step": 10077 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1519, + "step": 10078 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1639, + "step": 10079 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1554, + "step": 10080 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1469, + "step": 10081 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1895, + "step": 10082 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 10083 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1602, + "step": 10084 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 10085 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.201, + "step": 10086 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.2138, + "step": 10087 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1766, + "step": 10088 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1714, + "step": 10089 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1701, + "step": 10090 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1469, + "step": 10091 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1625, + "step": 10092 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1521, + "step": 10093 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1561, + "step": 10094 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1762, + "step": 10095 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1567, + "step": 10096 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1868, + "step": 10097 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.2073, + "step": 10098 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.2071, + "step": 10099 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1645, + "step": 10100 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 10101 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.2313, + "step": 10102 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 10103 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1538, + "step": 10104 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1624, + "step": 10105 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1687, + "step": 10106 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1998, + "step": 10107 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1894, + "step": 10108 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1992, + "step": 10109 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1943, + "step": 10110 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.161, + "step": 10111 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1815, + "step": 10112 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1651, + "step": 10113 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1611, + "step": 10114 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1626, + "step": 10115 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 10116 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1518, + "step": 10117 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 10118 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1916, + "step": 10119 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1958, + "step": 10120 + }, + { + "epoch": 0.61, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 10121 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.183, + "step": 10122 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 10123 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1997, + "step": 10124 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.2026, + "step": 10125 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 10126 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1833, + "step": 10127 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1686, + "step": 10128 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.225, + "step": 10129 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1697, + "step": 10130 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.2149, + "step": 10131 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1723, + "step": 10132 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1624, + "step": 10133 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 10134 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1807, + "step": 10135 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 10136 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1921, + "step": 10137 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1765, + "step": 10138 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 10139 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1645, + "step": 10140 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.2195, + "step": 10141 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1698, + "step": 10142 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 10143 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1809, + "step": 10144 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1769, + "step": 10145 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.2001, + "step": 10146 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1899, + "step": 10147 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 10148 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 10149 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 10150 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1567, + "step": 10151 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 10152 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1726, + "step": 10153 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 10154 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1712, + "step": 10155 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 10156 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1633, + "step": 10157 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1653, + "step": 10158 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.2149, + "step": 10159 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.2061, + "step": 10160 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1621, + "step": 10161 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1611, + "step": 10162 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1761, + "step": 10163 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 10164 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1715, + "step": 10165 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.2112, + "step": 10166 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1683, + "step": 10167 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 10168 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1657, + "step": 10169 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 10170 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1912, + "step": 10171 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1528, + "step": 10172 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1702, + "step": 10173 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1568, + "step": 10174 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1734, + "step": 10175 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 10176 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.2075, + "step": 10177 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1704, + "step": 10178 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 10179 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.2053, + "step": 10180 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1684, + "step": 10181 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1532, + "step": 10182 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1485, + "step": 10183 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.2074, + "step": 10184 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 10185 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1531, + "step": 10186 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1884, + "step": 10187 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1801, + "step": 10188 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 10189 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1918, + "step": 10190 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1555, + "step": 10191 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 10192 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 10193 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 10194 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 10195 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.159, + "step": 10196 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 10197 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1591, + "step": 10198 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 10199 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.162, + "step": 10200 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.2012, + "step": 10201 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 10202 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1924, + "step": 10203 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 10204 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 10205 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 10206 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1899, + "step": 10207 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 10208 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 10209 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1465, + "step": 10210 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1598, + "step": 10211 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 10212 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 10213 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 10214 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1687, + "step": 10215 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 10216 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 10217 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1953, + "step": 10218 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 10219 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 10220 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1868, + "step": 10221 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1629, + "step": 10222 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.163, + "step": 10223 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 10224 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1532, + "step": 10225 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1502, + "step": 10226 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1673, + "step": 10227 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 10228 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1706, + "step": 10229 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1583, + "step": 10230 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.199, + "step": 10231 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1592, + "step": 10232 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.2128, + "step": 10233 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 10234 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.2392, + "step": 10235 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 10236 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1796, + "step": 10237 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1705, + "step": 10238 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.158, + "step": 10239 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 10240 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1825, + "step": 10241 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 10242 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1621, + "step": 10243 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1607, + "step": 10244 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.165, + "step": 10245 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.2143, + "step": 10246 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 10247 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.2202, + "step": 10248 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 10249 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1947, + "step": 10250 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1836, + "step": 10251 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.174, + "step": 10252 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.173, + "step": 10253 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1886, + "step": 10254 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 10255 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 10256 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1707, + "step": 10257 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 10258 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 10259 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1689, + "step": 10260 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 10261 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 10262 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1807, + "step": 10263 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1499, + "step": 10264 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 10265 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 10266 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 10267 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 10268 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1692, + "step": 10269 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.196, + "step": 10270 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1959, + "step": 10271 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 10272 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1965, + "step": 10273 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.217, + "step": 10274 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1682, + "step": 10275 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1553, + "step": 10276 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1672, + "step": 10277 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 10278 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 10279 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1986, + "step": 10280 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 10281 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1886, + "step": 10282 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1685, + "step": 10283 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.1678, + "step": 10284 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.152, + "step": 10285 + }, + { + "epoch": 0.62, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 10286 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1551, + "step": 10287 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1873, + "step": 10288 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.2051, + "step": 10289 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.2199, + "step": 10290 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1624, + "step": 10291 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 10292 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 10293 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.215, + "step": 10294 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1755, + "step": 10295 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1583, + "step": 10296 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.2155, + "step": 10297 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1566, + "step": 10298 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1964, + "step": 10299 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1939, + "step": 10300 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1838, + "step": 10301 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 10302 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1847, + "step": 10303 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 10304 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1779, + "step": 10305 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.154, + "step": 10306 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1944, + "step": 10307 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1733, + "step": 10308 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 10309 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 10310 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.2014, + "step": 10311 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1847, + "step": 10312 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.2031, + "step": 10313 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1783, + "step": 10314 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1627, + "step": 10315 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1687, + "step": 10316 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1662, + "step": 10317 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1607, + "step": 10318 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.2001, + "step": 10319 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.2094, + "step": 10320 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 10321 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.203, + "step": 10322 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1655, + "step": 10323 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1858, + "step": 10324 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1624, + "step": 10325 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1511, + "step": 10326 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1658, + "step": 10327 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 10328 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 10329 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.198, + "step": 10330 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1743, + "step": 10331 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 10332 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1798, + "step": 10333 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1557, + "step": 10334 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1623, + "step": 10335 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1993, + "step": 10336 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1633, + "step": 10337 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 10338 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1687, + "step": 10339 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1852, + "step": 10340 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1583, + "step": 10341 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.165, + "step": 10342 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 10343 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1692, + "step": 10344 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 10345 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1732, + "step": 10346 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.2294, + "step": 10347 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1655, + "step": 10348 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1841, + "step": 10349 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.179, + "step": 10350 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1665, + "step": 10351 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.2039, + "step": 10352 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1607, + "step": 10353 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1728, + "step": 10354 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.164, + "step": 10355 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 10356 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1642, + "step": 10357 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1891, + "step": 10358 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 10359 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1906, + "step": 10360 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 10361 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 10362 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1666, + "step": 10363 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.167, + "step": 10364 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1857, + "step": 10365 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 10366 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 10367 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1842, + "step": 10368 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1516, + "step": 10369 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1767, + "step": 10370 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1906, + "step": 10371 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 10372 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1582, + "step": 10373 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 10374 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1706, + "step": 10375 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.185, + "step": 10376 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1717, + "step": 10377 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 10378 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 10379 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 10380 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1952, + "step": 10381 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 10382 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.2076, + "step": 10383 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1859, + "step": 10384 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.0855, + "step": 10385 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.2161, + "step": 10386 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 10387 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 10388 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1839, + "step": 10389 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.161, + "step": 10390 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.2249, + "step": 10391 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 10392 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 10393 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 10394 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1912, + "step": 10395 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 10396 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 10397 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 10398 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1769, + "step": 10399 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1623, + "step": 10400 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1525, + "step": 10401 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1983, + "step": 10402 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1838, + "step": 10403 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.187, + "step": 10404 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 10405 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1478, + "step": 10406 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1672, + "step": 10407 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1759, + "step": 10408 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 10409 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1596, + "step": 10410 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 10411 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1866, + "step": 10412 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 10413 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1782, + "step": 10414 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1677, + "step": 10415 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1447, + "step": 10416 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1964, + "step": 10417 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1673, + "step": 10418 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.2072, + "step": 10419 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.2127, + "step": 10420 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.2316, + "step": 10421 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.179, + "step": 10422 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 10423 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 10424 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1878, + "step": 10425 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1612, + "step": 10426 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 10427 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1805, + "step": 10428 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 10429 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1716, + "step": 10430 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 10431 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1956, + "step": 10432 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 10433 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 10434 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 10435 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.2115, + "step": 10436 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 10437 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1747, + "step": 10438 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1886, + "step": 10439 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1775, + "step": 10440 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 10441 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 10442 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1934, + "step": 10443 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 10444 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.168, + "step": 10445 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 10446 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1645, + "step": 10447 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1598, + "step": 10448 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1815, + "step": 10449 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1985, + "step": 10450 + }, + { + "epoch": 0.63, + "learning_rate": 2e-05, + "loss": 0.1632, + "step": 10451 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1691, + "step": 10452 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.2079, + "step": 10453 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 10454 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1708, + "step": 10455 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 10456 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.164, + "step": 10457 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1675, + "step": 10458 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.2074, + "step": 10459 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.167, + "step": 10460 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 10461 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1689, + "step": 10462 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 10463 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 10464 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1734, + "step": 10465 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1705, + "step": 10466 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 10467 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 10468 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 10469 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1689, + "step": 10470 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 10471 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1998, + "step": 10472 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 10473 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 10474 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1873, + "step": 10475 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1517, + "step": 10476 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1476, + "step": 10477 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.2063, + "step": 10478 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 10479 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 10480 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1679, + "step": 10481 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.2157, + "step": 10482 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1928, + "step": 10483 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1649, + "step": 10484 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1883, + "step": 10485 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1851, + "step": 10486 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1621, + "step": 10487 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 10488 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1866, + "step": 10489 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 10490 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.158, + "step": 10491 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 10492 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 10493 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.185, + "step": 10494 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 10495 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1644, + "step": 10496 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 10497 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1596, + "step": 10498 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1777, + "step": 10499 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1666, + "step": 10500 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 10501 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1675, + "step": 10502 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1996, + "step": 10503 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 10504 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1686, + "step": 10505 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1573, + "step": 10506 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1651, + "step": 10507 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1538, + "step": 10508 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1997, + "step": 10509 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1843, + "step": 10510 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1532, + "step": 10511 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1811, + "step": 10512 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.2028, + "step": 10513 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1666, + "step": 10514 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 10515 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1963, + "step": 10516 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.2005, + "step": 10517 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.156, + "step": 10518 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 10519 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1877, + "step": 10520 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1705, + "step": 10521 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1618, + "step": 10522 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.2021, + "step": 10523 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1456, + "step": 10524 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.2162, + "step": 10525 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1875, + "step": 10526 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 10527 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1752, + "step": 10528 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1772, + "step": 10529 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1595, + "step": 10530 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 10531 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1866, + "step": 10532 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.2217, + "step": 10533 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 10534 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1898, + "step": 10535 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1832, + "step": 10536 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 10537 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1794, + "step": 10538 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1807, + "step": 10539 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1929, + "step": 10540 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1667, + "step": 10541 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 10542 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.2061, + "step": 10543 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1637, + "step": 10544 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 10545 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1786, + "step": 10546 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1666, + "step": 10547 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 10548 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1637, + "step": 10549 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.2207, + "step": 10550 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1629, + "step": 10551 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1558, + "step": 10552 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 10553 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 10554 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1664, + "step": 10555 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 10556 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1498, + "step": 10557 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 10558 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.152, + "step": 10559 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1552, + "step": 10560 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.16, + "step": 10561 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 10562 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 10563 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.161, + "step": 10564 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 10565 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1585, + "step": 10566 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 10567 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 10568 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1478, + "step": 10569 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1499, + "step": 10570 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1495, + "step": 10571 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1977, + "step": 10572 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1775, + "step": 10573 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.159, + "step": 10574 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1603, + "step": 10575 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1846, + "step": 10576 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1962, + "step": 10577 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 10578 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.161, + "step": 10579 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1698, + "step": 10580 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 10581 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1856, + "step": 10582 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1837, + "step": 10583 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 10584 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 10585 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1844, + "step": 10586 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1646, + "step": 10587 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 10588 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1896, + "step": 10589 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1728, + "step": 10590 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.167, + "step": 10591 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 10592 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.156, + "step": 10593 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1681, + "step": 10594 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1713, + "step": 10595 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 10596 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1558, + "step": 10597 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1655, + "step": 10598 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1907, + "step": 10599 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 10600 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.2025, + "step": 10601 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1746, + "step": 10602 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 10603 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.156, + "step": 10604 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1961, + "step": 10605 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1593, + "step": 10606 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.144, + "step": 10607 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 10608 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.2113, + "step": 10609 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 10610 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.16, + "step": 10611 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1591, + "step": 10612 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 10613 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1554, + "step": 10614 + }, + { + "epoch": 0.64, + "learning_rate": 2e-05, + "loss": 0.1969, + "step": 10615 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1673, + "step": 10616 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 10617 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 10618 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.158, + "step": 10619 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1771, + "step": 10620 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1503, + "step": 10621 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1893, + "step": 10622 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1935, + "step": 10623 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1645, + "step": 10624 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1563, + "step": 10625 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1996, + "step": 10626 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1542, + "step": 10627 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1939, + "step": 10628 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1625, + "step": 10629 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1687, + "step": 10630 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 10631 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1875, + "step": 10632 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.165, + "step": 10633 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 10634 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.2531, + "step": 10635 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.2163, + "step": 10636 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 10637 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 10638 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1668, + "step": 10639 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.2172, + "step": 10640 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1645, + "step": 10641 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.148, + "step": 10642 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 10643 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.183, + "step": 10644 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1892, + "step": 10645 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.2075, + "step": 10646 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1762, + "step": 10647 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1495, + "step": 10648 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 10649 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 10650 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1796, + "step": 10651 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1552, + "step": 10652 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1649, + "step": 10653 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1792, + "step": 10654 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.2469, + "step": 10655 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 10656 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 10657 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 10658 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1601, + "step": 10659 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1936, + "step": 10660 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1915, + "step": 10661 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 10662 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 10663 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1681, + "step": 10664 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1824, + "step": 10665 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1685, + "step": 10666 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 10667 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 10668 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1814, + "step": 10669 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1606, + "step": 10670 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 10671 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 10672 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 10673 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1669, + "step": 10674 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1978, + "step": 10675 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1715, + "step": 10676 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1714, + "step": 10677 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1709, + "step": 10678 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1813, + "step": 10679 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1574, + "step": 10680 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1645, + "step": 10681 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1647, + "step": 10682 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1792, + "step": 10683 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 10684 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 10685 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.2143, + "step": 10686 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 10687 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1878, + "step": 10688 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1969, + "step": 10689 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 10690 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1834, + "step": 10691 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.2113, + "step": 10692 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 10693 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 10694 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1608, + "step": 10695 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 10696 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.169, + "step": 10697 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 10698 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1724, + "step": 10699 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.208, + "step": 10700 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.2153, + "step": 10701 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1751, + "step": 10702 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 10703 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 10704 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.2096, + "step": 10705 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1708, + "step": 10706 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1845, + "step": 10707 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1574, + "step": 10708 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 10709 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1625, + "step": 10710 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1521, + "step": 10711 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 10712 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1769, + "step": 10713 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.168, + "step": 10714 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.178, + "step": 10715 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 10716 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1624, + "step": 10717 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1593, + "step": 10718 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1692, + "step": 10719 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1921, + "step": 10720 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 10721 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 10722 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1667, + "step": 10723 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 10724 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.2046, + "step": 10725 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1648, + "step": 10726 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.16, + "step": 10727 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 10728 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1829, + "step": 10729 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 10730 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1881, + "step": 10731 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 10732 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1807, + "step": 10733 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 10734 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1741, + "step": 10735 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1737, + "step": 10736 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 10737 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 10738 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1651, + "step": 10739 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.184, + "step": 10740 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 10741 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1777, + "step": 10742 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1988, + "step": 10743 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.2191, + "step": 10744 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1776, + "step": 10745 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 10746 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 10747 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1845, + "step": 10748 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1691, + "step": 10749 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 10750 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 10751 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1786, + "step": 10752 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1692, + "step": 10753 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 10754 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1652, + "step": 10755 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 10756 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.144, + "step": 10757 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1551, + "step": 10758 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1773, + "step": 10759 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1662, + "step": 10760 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 10761 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1825, + "step": 10762 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 10763 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1621, + "step": 10764 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1711, + "step": 10765 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 10766 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 10767 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 10768 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1843, + "step": 10769 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1676, + "step": 10770 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1531, + "step": 10771 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1785, + "step": 10772 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 10773 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1606, + "step": 10774 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1807, + "step": 10775 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 10776 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.168, + "step": 10777 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1904, + "step": 10778 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1807, + "step": 10779 + }, + { + "epoch": 0.65, + "learning_rate": 2e-05, + "loss": 0.1804, + "step": 10780 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1596, + "step": 10781 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1624, + "step": 10782 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1592, + "step": 10783 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.2066, + "step": 10784 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1578, + "step": 10785 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 10786 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 10787 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1661, + "step": 10788 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1787, + "step": 10789 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.196, + "step": 10790 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 10791 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1704, + "step": 10792 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1513, + "step": 10793 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1817, + "step": 10794 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1526, + "step": 10795 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1742, + "step": 10796 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1992, + "step": 10797 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1595, + "step": 10798 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 10799 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1864, + "step": 10800 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1775, + "step": 10801 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1655, + "step": 10802 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1895, + "step": 10803 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 10804 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1746, + "step": 10805 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1624, + "step": 10806 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1809, + "step": 10807 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1979, + "step": 10808 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 10809 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.162, + "step": 10810 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 10811 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1786, + "step": 10812 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 10813 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1635, + "step": 10814 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1675, + "step": 10815 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1818, + "step": 10816 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 10817 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1955, + "step": 10818 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.177, + "step": 10819 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1842, + "step": 10820 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1635, + "step": 10821 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1648, + "step": 10822 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1801, + "step": 10823 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.2213, + "step": 10824 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1884, + "step": 10825 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1607, + "step": 10826 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1772, + "step": 10827 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 10828 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.2123, + "step": 10829 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1531, + "step": 10830 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1459, + "step": 10831 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.2098, + "step": 10832 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 10833 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1631, + "step": 10834 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 10835 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.2205, + "step": 10836 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.2081, + "step": 10837 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1937, + "step": 10838 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1976, + "step": 10839 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.2079, + "step": 10840 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 10841 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1918, + "step": 10842 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 10843 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1718, + "step": 10844 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.2033, + "step": 10845 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.201, + "step": 10846 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 10847 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1747, + "step": 10848 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.2, + "step": 10849 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1583, + "step": 10850 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 10851 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 10852 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1886, + "step": 10853 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1744, + "step": 10854 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 10855 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 10856 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1733, + "step": 10857 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 10858 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 10859 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1658, + "step": 10860 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1854, + "step": 10861 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1867, + "step": 10862 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1888, + "step": 10863 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1724, + "step": 10864 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1786, + "step": 10865 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1602, + "step": 10866 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 10867 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1655, + "step": 10868 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 10869 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 10870 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1873, + "step": 10871 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1833, + "step": 10872 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 10873 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1768, + "step": 10874 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1839, + "step": 10875 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 10876 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1838, + "step": 10877 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1629, + "step": 10878 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 10879 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 10880 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 10881 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.16, + "step": 10882 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.179, + "step": 10883 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.2072, + "step": 10884 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1834, + "step": 10885 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 10886 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.2149, + "step": 10887 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.182, + "step": 10888 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1904, + "step": 10889 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.156, + "step": 10890 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.2493, + "step": 10891 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1518, + "step": 10892 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.168, + "step": 10893 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1742, + "step": 10894 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1711, + "step": 10895 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1573, + "step": 10896 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1599, + "step": 10897 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.174, + "step": 10898 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1994, + "step": 10899 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1796, + "step": 10900 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1599, + "step": 10901 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1907, + "step": 10902 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1654, + "step": 10903 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1792, + "step": 10904 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1898, + "step": 10905 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1951, + "step": 10906 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 10907 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1812, + "step": 10908 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1669, + "step": 10909 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1553, + "step": 10910 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.219, + "step": 10911 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 10912 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1977, + "step": 10913 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 10914 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1586, + "step": 10915 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1667, + "step": 10916 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1887, + "step": 10917 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1698, + "step": 10918 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.184, + "step": 10919 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1897, + "step": 10920 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 10921 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1877, + "step": 10922 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.152, + "step": 10923 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1524, + "step": 10924 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1593, + "step": 10925 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1854, + "step": 10926 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1949, + "step": 10927 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.17, + "step": 10928 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 10929 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1989, + "step": 10930 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1641, + "step": 10931 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.0952, + "step": 10932 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1822, + "step": 10933 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1769, + "step": 10934 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1627, + "step": 10935 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.174, + "step": 10936 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.2019, + "step": 10937 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 10938 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.191, + "step": 10939 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1623, + "step": 10940 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 10941 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 10942 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 10943 + }, + { + "epoch": 0.66, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 10944 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.2375, + "step": 10945 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.172, + "step": 10946 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1771, + "step": 10947 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1574, + "step": 10948 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1585, + "step": 10949 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1608, + "step": 10950 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 10951 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1531, + "step": 10952 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1814, + "step": 10953 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1799, + "step": 10954 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.2032, + "step": 10955 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1596, + "step": 10956 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 10957 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1759, + "step": 10958 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.2142, + "step": 10959 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1766, + "step": 10960 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 10961 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1628, + "step": 10962 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1702, + "step": 10963 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1716, + "step": 10964 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1765, + "step": 10965 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 10966 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1861, + "step": 10967 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 10968 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 10969 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.158, + "step": 10970 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1586, + "step": 10971 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1713, + "step": 10972 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 10973 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.2007, + "step": 10974 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1731, + "step": 10975 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1725, + "step": 10976 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1841, + "step": 10977 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1595, + "step": 10978 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.2375, + "step": 10979 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1637, + "step": 10980 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1951, + "step": 10981 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1835, + "step": 10982 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 10983 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 10984 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 10985 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.2265, + "step": 10986 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 10987 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 10988 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 10989 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.2172, + "step": 10990 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1809, + "step": 10991 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1561, + "step": 10992 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1442, + "step": 10993 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1785, + "step": 10994 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.2138, + "step": 10995 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1835, + "step": 10996 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 10997 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 10998 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 10999 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1958, + "step": 11000 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 11001 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 11002 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 11003 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1941, + "step": 11004 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1875, + "step": 11005 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.164, + "step": 11006 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1761, + "step": 11007 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1953, + "step": 11008 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 11009 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1805, + "step": 11010 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1485, + "step": 11011 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1564, + "step": 11012 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1935, + "step": 11013 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 11014 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1645, + "step": 11015 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1662, + "step": 11016 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 11017 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1775, + "step": 11018 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 11019 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1683, + "step": 11020 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1863, + "step": 11021 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.144, + "step": 11022 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1878, + "step": 11023 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1814, + "step": 11024 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.164, + "step": 11025 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 11026 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 11027 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1601, + "step": 11028 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 11029 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 11030 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1465, + "step": 11031 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1784, + "step": 11032 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 11033 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.2254, + "step": 11034 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1645, + "step": 11035 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1593, + "step": 11036 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1703, + "step": 11037 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1897, + "step": 11038 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1541, + "step": 11039 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 11040 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 11041 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 11042 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1604, + "step": 11043 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1965, + "step": 11044 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1588, + "step": 11045 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1478, + "step": 11046 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1985, + "step": 11047 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 11048 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 11049 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.218, + "step": 11050 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.178, + "step": 11051 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1675, + "step": 11052 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 11053 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.2217, + "step": 11054 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 11055 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 11056 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1532, + "step": 11057 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.168, + "step": 11058 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 11059 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 11060 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 11061 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1954, + "step": 11062 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 11063 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1538, + "step": 11064 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 11065 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 11066 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1645, + "step": 11067 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1568, + "step": 11068 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 11069 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1608, + "step": 11070 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.2007, + "step": 11071 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1579, + "step": 11072 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 11073 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 11074 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.194, + "step": 11075 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1634, + "step": 11076 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1561, + "step": 11077 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 11078 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 11079 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1789, + "step": 11080 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 11081 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1749, + "step": 11082 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.2142, + "step": 11083 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1577, + "step": 11084 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 11085 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.169, + "step": 11086 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.163, + "step": 11087 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1683, + "step": 11088 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1725, + "step": 11089 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 11090 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.193, + "step": 11091 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1767, + "step": 11092 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 11093 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 11094 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1649, + "step": 11095 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 11096 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1796, + "step": 11097 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 11098 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1832, + "step": 11099 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.155, + "step": 11100 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1589, + "step": 11101 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 11102 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1746, + "step": 11103 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1658, + "step": 11104 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1731, + "step": 11105 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1883, + "step": 11106 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 11107 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 11108 + }, + { + "epoch": 0.67, + "learning_rate": 2e-05, + "loss": 0.1453, + "step": 11109 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1683, + "step": 11110 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1462, + "step": 11111 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1607, + "step": 11112 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 11113 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1773, + "step": 11114 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1664, + "step": 11115 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1746, + "step": 11116 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1604, + "step": 11117 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1598, + "step": 11118 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1661, + "step": 11119 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1663, + "step": 11120 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 11121 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.2135, + "step": 11122 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1576, + "step": 11123 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1955, + "step": 11124 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1795, + "step": 11125 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1748, + "step": 11126 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 11127 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1908, + "step": 11128 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 11129 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 11130 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1838, + "step": 11131 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 11132 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1625, + "step": 11133 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.2067, + "step": 11134 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1928, + "step": 11135 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 11136 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 11137 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 11138 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 11139 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1552, + "step": 11140 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1534, + "step": 11141 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1966, + "step": 11142 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1672, + "step": 11143 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 11144 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 11145 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1603, + "step": 11146 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.2018, + "step": 11147 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1705, + "step": 11148 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1621, + "step": 11149 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1657, + "step": 11150 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.2018, + "step": 11151 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1811, + "step": 11152 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.2083, + "step": 11153 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1725, + "step": 11154 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 11155 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1776, + "step": 11156 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1759, + "step": 11157 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1842, + "step": 11158 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 11159 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 11160 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1808, + "step": 11161 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1554, + "step": 11162 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.177, + "step": 11163 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1975, + "step": 11164 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 11165 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1837, + "step": 11166 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1764, + "step": 11167 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 11168 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1865, + "step": 11169 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.2147, + "step": 11170 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1787, + "step": 11171 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.2157, + "step": 11172 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 11173 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 11174 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1718, + "step": 11175 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.2027, + "step": 11176 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1598, + "step": 11177 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 11178 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1499, + "step": 11179 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 11180 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1456, + "step": 11181 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.2197, + "step": 11182 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.2109, + "step": 11183 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 11184 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1672, + "step": 11185 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 11186 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 11187 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1645, + "step": 11188 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1932, + "step": 11189 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 11190 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1677, + "step": 11191 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1991, + "step": 11192 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1915, + "step": 11193 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1805, + "step": 11194 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 11195 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1841, + "step": 11196 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1841, + "step": 11197 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1987, + "step": 11198 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1956, + "step": 11199 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 11200 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1716, + "step": 11201 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 11202 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 11203 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1805, + "step": 11204 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1601, + "step": 11205 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 11206 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 11207 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 11208 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1453, + "step": 11209 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.184, + "step": 11210 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 11211 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 11212 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1606, + "step": 11213 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 11214 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1734, + "step": 11215 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1795, + "step": 11216 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1687, + "step": 11217 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.2215, + "step": 11218 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1867, + "step": 11219 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1825, + "step": 11220 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.2104, + "step": 11221 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 11222 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1666, + "step": 11223 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 11224 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.182, + "step": 11225 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1646, + "step": 11226 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1442, + "step": 11227 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1828, + "step": 11228 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1531, + "step": 11229 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 11230 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1656, + "step": 11231 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 11232 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1495, + "step": 11233 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1736, + "step": 11234 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1696, + "step": 11235 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 11236 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.17, + "step": 11237 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.197, + "step": 11238 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1815, + "step": 11239 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 11240 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 11241 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1697, + "step": 11242 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 11243 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 11244 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1488, + "step": 11245 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 11246 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1832, + "step": 11247 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1826, + "step": 11248 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 11249 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1593, + "step": 11250 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.2291, + "step": 11251 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1538, + "step": 11252 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 11253 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1816, + "step": 11254 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1755, + "step": 11255 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1568, + "step": 11256 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 11257 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1915, + "step": 11258 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1719, + "step": 11259 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1937, + "step": 11260 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 11261 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 11262 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.211, + "step": 11263 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.2123, + "step": 11264 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 11265 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.2397, + "step": 11266 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1841, + "step": 11267 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1613, + "step": 11268 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 11269 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 11270 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 11271 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 11272 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.2027, + "step": 11273 + }, + { + "epoch": 0.68, + "learning_rate": 2e-05, + "loss": 0.1911, + "step": 11274 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1764, + "step": 11275 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1656, + "step": 11276 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1586, + "step": 11277 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 11278 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 11279 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1616, + "step": 11280 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 11281 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1757, + "step": 11282 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 11283 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1915, + "step": 11284 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1611, + "step": 11285 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1696, + "step": 11286 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1905, + "step": 11287 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.2176, + "step": 11288 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1892, + "step": 11289 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1884, + "step": 11290 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 11291 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1783, + "step": 11292 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 11293 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1687, + "step": 11294 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 11295 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1941, + "step": 11296 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1687, + "step": 11297 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1583, + "step": 11298 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.2005, + "step": 11299 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1749, + "step": 11300 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1515, + "step": 11301 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 11302 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 11303 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 11304 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.2128, + "step": 11305 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1783, + "step": 11306 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1816, + "step": 11307 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1848, + "step": 11308 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1918, + "step": 11309 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1798, + "step": 11310 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.2007, + "step": 11311 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.2092, + "step": 11312 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 11313 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1738, + "step": 11314 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 11315 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1816, + "step": 11316 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1597, + "step": 11317 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 11318 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 11319 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 11320 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1847, + "step": 11321 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1599, + "step": 11322 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1709, + "step": 11323 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1622, + "step": 11324 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1862, + "step": 11325 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 11326 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 11327 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1773, + "step": 11328 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 11329 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1708, + "step": 11330 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.2005, + "step": 11331 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1626, + "step": 11332 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1704, + "step": 11333 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1732, + "step": 11334 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.2066, + "step": 11335 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 11336 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 11337 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 11338 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1794, + "step": 11339 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.199, + "step": 11340 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.2234, + "step": 11341 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 11342 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 11343 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 11344 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 11345 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1671, + "step": 11346 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 11347 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1616, + "step": 11348 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 11349 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1914, + "step": 11350 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1645, + "step": 11351 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.2363, + "step": 11352 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1666, + "step": 11353 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 11354 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 11355 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1981, + "step": 11356 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1585, + "step": 11357 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 11358 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.2128, + "step": 11359 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 11360 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1779, + "step": 11361 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.2076, + "step": 11362 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1688, + "step": 11363 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1643, + "step": 11364 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 11365 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 11366 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 11367 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1794, + "step": 11368 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1998, + "step": 11369 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1694, + "step": 11370 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1684, + "step": 11371 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1752, + "step": 11372 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1734, + "step": 11373 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1733, + "step": 11374 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.2011, + "step": 11375 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 11376 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 11377 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1577, + "step": 11378 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1593, + "step": 11379 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1821, + "step": 11380 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1617, + "step": 11381 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 11382 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.155, + "step": 11383 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 11384 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 11385 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 11386 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.161, + "step": 11387 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1661, + "step": 11388 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 11389 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.206, + "step": 11390 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1686, + "step": 11391 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1994, + "step": 11392 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1679, + "step": 11393 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 11394 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 11395 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 11396 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 11397 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1748, + "step": 11398 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 11399 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.167, + "step": 11400 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1608, + "step": 11401 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1607, + "step": 11402 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1899, + "step": 11403 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 11404 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 11405 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1605, + "step": 11406 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1976, + "step": 11407 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.172, + "step": 11408 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.2072, + "step": 11409 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 11410 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 11411 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1924, + "step": 11412 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1748, + "step": 11413 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 11414 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 11415 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1666, + "step": 11416 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 11417 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1662, + "step": 11418 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1687, + "step": 11419 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 11420 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 11421 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 11422 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.2011, + "step": 11423 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1808, + "step": 11424 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 11425 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.146, + "step": 11426 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1654, + "step": 11427 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 11428 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1954, + "step": 11429 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1678, + "step": 11430 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1898, + "step": 11431 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1577, + "step": 11432 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1862, + "step": 11433 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 11434 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1453, + "step": 11435 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1742, + "step": 11436 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1827, + "step": 11437 + }, + { + "epoch": 0.69, + "learning_rate": 2e-05, + "loss": 0.1578, + "step": 11438 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 11439 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 11440 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 11441 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1629, + "step": 11442 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1876, + "step": 11443 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 11444 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1813, + "step": 11445 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1561, + "step": 11446 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1968, + "step": 11447 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1762, + "step": 11448 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 11449 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 11450 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1765, + "step": 11451 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 11452 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1564, + "step": 11453 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1826, + "step": 11454 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1567, + "step": 11455 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 11456 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1713, + "step": 11457 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.2128, + "step": 11458 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 11459 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 11460 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1673, + "step": 11461 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 11462 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1751, + "step": 11463 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 11464 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.2197, + "step": 11465 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 11466 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1511, + "step": 11467 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.186, + "step": 11468 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 11469 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 11470 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.2259, + "step": 11471 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1652, + "step": 11472 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 11473 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1563, + "step": 11474 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1574, + "step": 11475 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1818, + "step": 11476 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1803, + "step": 11477 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1663, + "step": 11478 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.183, + "step": 11479 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1768, + "step": 11480 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 11481 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1701, + "step": 11482 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 11483 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1689, + "step": 11484 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1469, + "step": 11485 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.2069, + "step": 11486 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 11487 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1742, + "step": 11488 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1733, + "step": 11489 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.2091, + "step": 11490 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1926, + "step": 11491 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 11492 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1824, + "step": 11493 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1829, + "step": 11494 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.174, + "step": 11495 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1672, + "step": 11496 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.193, + "step": 11497 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 11498 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.176, + "step": 11499 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1972, + "step": 11500 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1686, + "step": 11501 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 11502 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.2205, + "step": 11503 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1685, + "step": 11504 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1738, + "step": 11505 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1682, + "step": 11506 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1696, + "step": 11507 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1841, + "step": 11508 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.172, + "step": 11509 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 11510 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.148, + "step": 11511 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1949, + "step": 11512 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1758, + "step": 11513 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.17, + "step": 11514 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1568, + "step": 11515 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 11516 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 11517 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1535, + "step": 11518 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1754, + "step": 11519 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1607, + "step": 11520 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1758, + "step": 11521 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 11522 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 11523 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.153, + "step": 11524 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.2109, + "step": 11525 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1857, + "step": 11526 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 11527 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 11528 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 11529 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1847, + "step": 11530 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.2309, + "step": 11531 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 11532 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1874, + "step": 11533 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1781, + "step": 11534 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1485, + "step": 11535 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 11536 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1513, + "step": 11537 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 11538 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 11539 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1741, + "step": 11540 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1533, + "step": 11541 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.158, + "step": 11542 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1748, + "step": 11543 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 11544 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1771, + "step": 11545 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 11546 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1538, + "step": 11547 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 11548 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1828, + "step": 11549 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1928, + "step": 11550 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.193, + "step": 11551 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1754, + "step": 11552 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1838, + "step": 11553 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 11554 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1526, + "step": 11555 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 11556 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1897, + "step": 11557 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 11558 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 11559 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 11560 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1638, + "step": 11561 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 11562 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1844, + "step": 11563 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1874, + "step": 11564 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1856, + "step": 11565 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1599, + "step": 11566 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 11567 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 11568 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1835, + "step": 11569 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.2273, + "step": 11570 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1616, + "step": 11571 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 11572 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1761, + "step": 11573 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1674, + "step": 11574 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1527, + "step": 11575 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 11576 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.193, + "step": 11577 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1593, + "step": 11578 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1666, + "step": 11579 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.188, + "step": 11580 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.2027, + "step": 11581 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 11582 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1573, + "step": 11583 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 11584 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 11585 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 11586 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1525, + "step": 11587 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1531, + "step": 11588 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 11589 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1748, + "step": 11590 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1617, + "step": 11591 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 11592 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1676, + "step": 11593 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 11594 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1796, + "step": 11595 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1787, + "step": 11596 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1859, + "step": 11597 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 11598 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 11599 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1886, + "step": 11600 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1947, + "step": 11601 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1613, + "step": 11602 + }, + { + "epoch": 0.7, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 11603 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1841, + "step": 11604 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1835, + "step": 11605 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 11606 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.2092, + "step": 11607 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1553, + "step": 11608 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1789, + "step": 11609 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1698, + "step": 11610 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 11611 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.2025, + "step": 11612 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1781, + "step": 11613 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1577, + "step": 11614 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1563, + "step": 11615 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1712, + "step": 11616 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 11617 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 11618 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1681, + "step": 11619 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 11620 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.2253, + "step": 11621 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1792, + "step": 11622 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 11623 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1898, + "step": 11624 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1685, + "step": 11625 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1639, + "step": 11626 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 11627 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1798, + "step": 11628 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 11629 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1699, + "step": 11630 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1505, + "step": 11631 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1866, + "step": 11632 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1499, + "step": 11633 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1846, + "step": 11634 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 11635 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1877, + "step": 11636 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1709, + "step": 11637 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1745, + "step": 11638 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.2041, + "step": 11639 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.2143, + "step": 11640 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1926, + "step": 11641 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 11642 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.2004, + "step": 11643 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1548, + "step": 11644 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1621, + "step": 11645 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1927, + "step": 11646 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1622, + "step": 11647 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1657, + "step": 11648 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1953, + "step": 11649 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 11650 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1633, + "step": 11651 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1429, + "step": 11652 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1835, + "step": 11653 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1646, + "step": 11654 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1898, + "step": 11655 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.2205, + "step": 11656 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1557, + "step": 11657 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.214, + "step": 11658 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 11659 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1533, + "step": 11660 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.178, + "step": 11661 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1908, + "step": 11662 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1453, + "step": 11663 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 11664 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 11665 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 11666 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1945, + "step": 11667 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1879, + "step": 11668 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 11669 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 11670 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 11671 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1827, + "step": 11672 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.2175, + "step": 11673 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 11674 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 11675 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1558, + "step": 11676 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1646, + "step": 11677 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 11678 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 11679 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1959, + "step": 11680 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1561, + "step": 11681 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1872, + "step": 11682 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1661, + "step": 11683 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 11684 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 11685 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1665, + "step": 11686 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1935, + "step": 11687 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1662, + "step": 11688 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1951, + "step": 11689 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 11690 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1778, + "step": 11691 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 11692 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1614, + "step": 11693 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.188, + "step": 11694 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1822, + "step": 11695 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 11696 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1884, + "step": 11697 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1552, + "step": 11698 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.2137, + "step": 11699 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.196, + "step": 11700 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1563, + "step": 11701 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 11702 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.2019, + "step": 11703 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 11704 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.2071, + "step": 11705 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.18, + "step": 11706 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 11707 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1986, + "step": 11708 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 11709 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 11710 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1638, + "step": 11711 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 11712 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 11713 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 11714 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1639, + "step": 11715 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1656, + "step": 11716 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1803, + "step": 11717 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.184, + "step": 11718 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1895, + "step": 11719 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1743, + "step": 11720 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.2071, + "step": 11721 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.2027, + "step": 11722 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1602, + "step": 11723 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1669, + "step": 11724 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.2084, + "step": 11725 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 11726 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 11727 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 11728 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.2257, + "step": 11729 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 11730 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1762, + "step": 11731 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1503, + "step": 11732 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1726, + "step": 11733 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1643, + "step": 11734 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.178, + "step": 11735 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 11736 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1823, + "step": 11737 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 11738 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1871, + "step": 11739 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 11740 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1681, + "step": 11741 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 11742 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.171, + "step": 11743 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 11744 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1723, + "step": 11745 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 11746 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1926, + "step": 11747 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1485, + "step": 11748 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1674, + "step": 11749 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.2002, + "step": 11750 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1582, + "step": 11751 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 11752 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1771, + "step": 11753 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1761, + "step": 11754 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 11755 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1853, + "step": 11756 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1813, + "step": 11757 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1705, + "step": 11758 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1679, + "step": 11759 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1979, + "step": 11760 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 11761 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 11762 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1851, + "step": 11763 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.2072, + "step": 11764 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1632, + "step": 11765 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.2026, + "step": 11766 + }, + { + "epoch": 0.71, + "learning_rate": 2e-05, + "loss": 0.1546, + "step": 11767 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 11768 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 11769 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1715, + "step": 11770 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 11771 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1927, + "step": 11772 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1787, + "step": 11773 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 11774 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1586, + "step": 11775 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 11776 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1807, + "step": 11777 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 11778 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.2029, + "step": 11779 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1939, + "step": 11780 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1693, + "step": 11781 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1725, + "step": 11782 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1677, + "step": 11783 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1902, + "step": 11784 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1714, + "step": 11785 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1681, + "step": 11786 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1955, + "step": 11787 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.156, + "step": 11788 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1861, + "step": 11789 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 11790 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 11791 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1783, + "step": 11792 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1737, + "step": 11793 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1774, + "step": 11794 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 11795 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 11796 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1518, + "step": 11797 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1658, + "step": 11798 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1743, + "step": 11799 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 11800 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1723, + "step": 11801 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 11802 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 11803 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1867, + "step": 11804 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1865, + "step": 11805 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.2169, + "step": 11806 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 11807 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1742, + "step": 11808 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 11809 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1485, + "step": 11810 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1633, + "step": 11811 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1685, + "step": 11812 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1443, + "step": 11813 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 11814 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 11815 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1651, + "step": 11816 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1641, + "step": 11817 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.209, + "step": 11818 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1751, + "step": 11819 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1827, + "step": 11820 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.2095, + "step": 11821 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1898, + "step": 11822 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 11823 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1752, + "step": 11824 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.2056, + "step": 11825 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 11826 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 11827 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1634, + "step": 11828 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1598, + "step": 11829 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 11830 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.177, + "step": 11831 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.208, + "step": 11832 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1731, + "step": 11833 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.2206, + "step": 11834 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1985, + "step": 11835 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 11836 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 11837 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1443, + "step": 11838 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1621, + "step": 11839 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1562, + "step": 11840 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1758, + "step": 11841 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.17, + "step": 11842 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.235, + "step": 11843 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 11844 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 11845 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1586, + "step": 11846 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 11847 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1518, + "step": 11848 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1469, + "step": 11849 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 11850 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1503, + "step": 11851 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1671, + "step": 11852 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 11853 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 11854 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1775, + "step": 11855 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1685, + "step": 11856 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 11857 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 11858 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.2071, + "step": 11859 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 11860 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 11861 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 11862 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1961, + "step": 11863 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 11864 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1721, + "step": 11865 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1755, + "step": 11866 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.2059, + "step": 11867 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.2406, + "step": 11868 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1817, + "step": 11869 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1844, + "step": 11870 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1605, + "step": 11871 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 11872 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1818, + "step": 11873 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1502, + "step": 11874 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1702, + "step": 11875 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 11876 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.172, + "step": 11877 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 11878 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 11879 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1805, + "step": 11880 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 11881 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1505, + "step": 11882 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1696, + "step": 11883 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 11884 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1761, + "step": 11885 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.168, + "step": 11886 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 11887 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1627, + "step": 11888 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1826, + "step": 11889 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 11890 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1923, + "step": 11891 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1782, + "step": 11892 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 11893 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1733, + "step": 11894 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1875, + "step": 11895 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 11896 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1589, + "step": 11897 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1712, + "step": 11898 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.2026, + "step": 11899 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1799, + "step": 11900 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 11901 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 11902 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1982, + "step": 11903 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1766, + "step": 11904 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 11905 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 11906 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 11907 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 11908 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 11909 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 11910 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 11911 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1639, + "step": 11912 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1945, + "step": 11913 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 11914 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 11915 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 11916 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1638, + "step": 11917 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1941, + "step": 11918 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.2051, + "step": 11919 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1798, + "step": 11920 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 11921 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1783, + "step": 11922 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 11923 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.168, + "step": 11924 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1551, + "step": 11925 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.2041, + "step": 11926 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1624, + "step": 11927 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1699, + "step": 11928 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.2007, + "step": 11929 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1519, + "step": 11930 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1952, + "step": 11931 + }, + { + "epoch": 0.72, + "learning_rate": 2e-05, + "loss": 0.1855, + "step": 11932 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1462, + "step": 11933 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1516, + "step": 11934 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1758, + "step": 11935 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 11936 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1953, + "step": 11937 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 11938 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 11939 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 11940 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1614, + "step": 11941 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1712, + "step": 11942 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 11943 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1822, + "step": 11944 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 11945 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 11946 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 11947 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 11948 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1567, + "step": 11949 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 11950 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 11951 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1867, + "step": 11952 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 11953 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1717, + "step": 11954 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 11955 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 11956 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1538, + "step": 11957 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.17, + "step": 11958 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 11959 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 11960 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 11961 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1616, + "step": 11962 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1753, + "step": 11963 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 11964 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.2395, + "step": 11965 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 11966 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.164, + "step": 11967 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.2027, + "step": 11968 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1832, + "step": 11969 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.177, + "step": 11970 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 11971 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1924, + "step": 11972 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 11973 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1775, + "step": 11974 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1859, + "step": 11975 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1702, + "step": 11976 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.171, + "step": 11977 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1797, + "step": 11978 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1672, + "step": 11979 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1865, + "step": 11980 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 11981 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 11982 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 11983 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.168, + "step": 11984 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 11985 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 11986 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.165, + "step": 11987 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 11988 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.2171, + "step": 11989 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1909, + "step": 11990 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 11991 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1551, + "step": 11992 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 11993 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 11994 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 11995 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1567, + "step": 11996 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1668, + "step": 11997 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1616, + "step": 11998 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1583, + "step": 11999 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1714, + "step": 12000 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 12001 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 12002 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1779, + "step": 12003 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1626, + "step": 12004 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 12005 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 12006 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1975, + "step": 12007 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 12008 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1577, + "step": 12009 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1832, + "step": 12010 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1557, + "step": 12011 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1443, + "step": 12012 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1827, + "step": 12013 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1616, + "step": 12014 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1714, + "step": 12015 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1511, + "step": 12016 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 12017 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1562, + "step": 12018 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 12019 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 12020 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 12021 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 12022 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 12023 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 12024 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 12025 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1602, + "step": 12026 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1611, + "step": 12027 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1651, + "step": 12028 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 12029 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 12030 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1769, + "step": 12031 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1558, + "step": 12032 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 12033 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 12034 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.162, + "step": 12035 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 12036 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1798, + "step": 12037 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1828, + "step": 12038 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 12039 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1727, + "step": 12040 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1933, + "step": 12041 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 12042 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 12043 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1929, + "step": 12044 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1733, + "step": 12045 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1862, + "step": 12046 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1685, + "step": 12047 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.2019, + "step": 12048 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1499, + "step": 12049 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 12050 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1915, + "step": 12051 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1782, + "step": 12052 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1841, + "step": 12053 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 12054 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 12055 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1871, + "step": 12056 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1738, + "step": 12057 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1883, + "step": 12058 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1739, + "step": 12059 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1698, + "step": 12060 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 12061 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.2095, + "step": 12062 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1773, + "step": 12063 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.168, + "step": 12064 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 12065 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1954, + "step": 12066 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1577, + "step": 12067 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1733, + "step": 12068 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 12069 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 12070 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 12071 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1709, + "step": 12072 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.189, + "step": 12073 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1814, + "step": 12074 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1476, + "step": 12075 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1699, + "step": 12076 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1987, + "step": 12077 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.193, + "step": 12078 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1635, + "step": 12079 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1912, + "step": 12080 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 12081 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1735, + "step": 12082 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 12083 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 12084 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 12085 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 12086 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 12087 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 12088 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.181, + "step": 12089 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 12090 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1627, + "step": 12091 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 12092 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 12093 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1526, + "step": 12094 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 12095 + }, + { + "epoch": 0.73, + "learning_rate": 2e-05, + "loss": 0.1766, + "step": 12096 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 12097 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 12098 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1799, + "step": 12099 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 12100 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 12101 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 12102 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 12103 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.2025, + "step": 12104 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1832, + "step": 12105 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1853, + "step": 12106 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.146, + "step": 12107 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.2324, + "step": 12108 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.162, + "step": 12109 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1662, + "step": 12110 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.2034, + "step": 12111 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.162, + "step": 12112 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1739, + "step": 12113 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 12114 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 12115 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1719, + "step": 12116 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1815, + "step": 12117 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.2159, + "step": 12118 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 12119 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1854, + "step": 12120 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.2119, + "step": 12121 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1832, + "step": 12122 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1678, + "step": 12123 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1753, + "step": 12124 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.147, + "step": 12125 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.153, + "step": 12126 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1708, + "step": 12127 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.2103, + "step": 12128 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1927, + "step": 12129 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 12130 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 12131 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1731, + "step": 12132 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1664, + "step": 12133 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1685, + "step": 12134 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1597, + "step": 12135 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1785, + "step": 12136 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 12137 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1855, + "step": 12138 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 12139 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1721, + "step": 12140 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 12141 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1738, + "step": 12142 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1669, + "step": 12143 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.2215, + "step": 12144 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 12145 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 12146 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.2152, + "step": 12147 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.2197, + "step": 12148 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 12149 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 12150 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 12151 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.2203, + "step": 12152 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1663, + "step": 12153 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 12154 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.23, + "step": 12155 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1784, + "step": 12156 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1855, + "step": 12157 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 12158 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1925, + "step": 12159 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1741, + "step": 12160 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 12161 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 12162 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.2079, + "step": 12163 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1663, + "step": 12164 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 12165 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 12166 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1469, + "step": 12167 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1531, + "step": 12168 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1557, + "step": 12169 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1702, + "step": 12170 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1505, + "step": 12171 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1638, + "step": 12172 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1813, + "step": 12173 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1485, + "step": 12174 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1892, + "step": 12175 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.2106, + "step": 12176 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1617, + "step": 12177 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1828, + "step": 12178 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 12179 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1691, + "step": 12180 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 12181 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 12182 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1773, + "step": 12183 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.161, + "step": 12184 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 12185 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 12186 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.2047, + "step": 12187 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1519, + "step": 12188 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 12189 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1716, + "step": 12190 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1866, + "step": 12191 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 12192 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 12193 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1813, + "step": 12194 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.2114, + "step": 12195 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1969, + "step": 12196 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1828, + "step": 12197 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 12198 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.2088, + "step": 12199 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 12200 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1635, + "step": 12201 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1726, + "step": 12202 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1548, + "step": 12203 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1558, + "step": 12204 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 12205 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1626, + "step": 12206 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1548, + "step": 12207 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1629, + "step": 12208 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1532, + "step": 12209 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 12210 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 12211 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1497, + "step": 12212 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 12213 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1611, + "step": 12214 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1658, + "step": 12215 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1532, + "step": 12216 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 12217 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 12218 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 12219 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 12220 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 12221 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 12222 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1943, + "step": 12223 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 12224 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1686, + "step": 12225 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1498, + "step": 12226 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 12227 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 12228 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.2127, + "step": 12229 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1719, + "step": 12230 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1764, + "step": 12231 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 12232 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 12233 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 12234 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1476, + "step": 12235 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 12236 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 12237 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 12238 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1971, + "step": 12239 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 12240 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1874, + "step": 12241 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 12242 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 12243 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 12244 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1502, + "step": 12245 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1728, + "step": 12246 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 12247 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 12248 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 12249 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 12250 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 12251 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1897, + "step": 12252 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1588, + "step": 12253 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 12254 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1456, + "step": 12255 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 12256 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1633, + "step": 12257 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.2008, + "step": 12258 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1973, + "step": 12259 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.161, + "step": 12260 + }, + { + "epoch": 0.74, + "learning_rate": 2e-05, + "loss": 0.1476, + "step": 12261 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1864, + "step": 12262 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1717, + "step": 12263 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 12264 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1917, + "step": 12265 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1765, + "step": 12266 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1443, + "step": 12267 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 12268 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 12269 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 12270 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 12271 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1599, + "step": 12272 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1442, + "step": 12273 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1932, + "step": 12274 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 12275 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 12276 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1541, + "step": 12277 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.202, + "step": 12278 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1683, + "step": 12279 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1842, + "step": 12280 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.201, + "step": 12281 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.154, + "step": 12282 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.2118, + "step": 12283 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1827, + "step": 12284 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 12285 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1769, + "step": 12286 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 12287 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1611, + "step": 12288 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1505, + "step": 12289 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1498, + "step": 12290 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 12291 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.211, + "step": 12292 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.2026, + "step": 12293 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.175, + "step": 12294 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 12295 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 12296 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 12297 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 12298 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.144, + "step": 12299 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 12300 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1677, + "step": 12301 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1671, + "step": 12302 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 12303 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1969, + "step": 12304 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.2012, + "step": 12305 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 12306 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1495, + "step": 12307 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.183, + "step": 12308 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1718, + "step": 12309 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1641, + "step": 12310 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1597, + "step": 12311 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1637, + "step": 12312 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 12313 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 12314 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1702, + "step": 12315 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1674, + "step": 12316 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 12317 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 12318 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.172, + "step": 12319 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 12320 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 12321 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 12322 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 12323 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1513, + "step": 12324 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 12325 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 12326 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 12327 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 12328 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 12329 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.2078, + "step": 12330 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1649, + "step": 12331 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 12332 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 12333 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1621, + "step": 12334 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1699, + "step": 12335 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 12336 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.155, + "step": 12337 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.2343, + "step": 12338 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.176, + "step": 12339 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1527, + "step": 12340 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1768, + "step": 12341 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1914, + "step": 12342 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1886, + "step": 12343 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1978, + "step": 12344 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1858, + "step": 12345 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1688, + "step": 12346 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 12347 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 12348 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 12349 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1588, + "step": 12350 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.2196, + "step": 12351 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 12352 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 12353 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.2021, + "step": 12354 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 12355 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1656, + "step": 12356 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 12357 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 12358 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1579, + "step": 12359 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1589, + "step": 12360 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1805, + "step": 12361 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.2028, + "step": 12362 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1811, + "step": 12363 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1784, + "step": 12364 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 12365 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1626, + "step": 12366 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 12367 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.2063, + "step": 12368 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.2141, + "step": 12369 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1456, + "step": 12370 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 12371 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1497, + "step": 12372 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1634, + "step": 12373 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 12374 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 12375 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1613, + "step": 12376 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.173, + "step": 12377 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1991, + "step": 12378 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1649, + "step": 12379 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1823, + "step": 12380 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 12381 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.165, + "step": 12382 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1553, + "step": 12383 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1629, + "step": 12384 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 12385 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 12386 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1832, + "step": 12387 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1744, + "step": 12388 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 12389 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.2022, + "step": 12390 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 12391 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 12392 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1794, + "step": 12393 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 12394 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 12395 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1935, + "step": 12396 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1854, + "step": 12397 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 12398 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 12399 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1498, + "step": 12400 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 12401 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 12402 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 12403 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 12404 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1991, + "step": 12405 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1734, + "step": 12406 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 12407 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1497, + "step": 12408 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1573, + "step": 12409 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1511, + "step": 12410 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.2112, + "step": 12411 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 12412 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.161, + "step": 12413 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1525, + "step": 12414 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1662, + "step": 12415 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 12416 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1771, + "step": 12417 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1992, + "step": 12418 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 12419 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1835, + "step": 12420 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 12421 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 12422 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 12423 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 12424 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 12425 + }, + { + "epoch": 0.75, + "learning_rate": 2e-05, + "loss": 0.1586, + "step": 12426 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 12427 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 12428 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1621, + "step": 12429 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1524, + "step": 12430 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1778, + "step": 12431 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1918, + "step": 12432 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1651, + "step": 12433 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 12434 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1635, + "step": 12435 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1765, + "step": 12436 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 12437 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.2125, + "step": 12438 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1728, + "step": 12439 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.217, + "step": 12440 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.19, + "step": 12441 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 12442 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 12443 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.2247, + "step": 12444 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1459, + "step": 12445 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.16, + "step": 12446 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 12447 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1601, + "step": 12448 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1773, + "step": 12449 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1602, + "step": 12450 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1476, + "step": 12451 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 12452 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 12453 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 12454 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 12455 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1823, + "step": 12456 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1623, + "step": 12457 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1756, + "step": 12458 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1657, + "step": 12459 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1749, + "step": 12460 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 12461 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1643, + "step": 12462 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1889, + "step": 12463 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.2128, + "step": 12464 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 12465 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1929, + "step": 12466 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1662, + "step": 12467 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 12468 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1554, + "step": 12469 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1582, + "step": 12470 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.2129, + "step": 12471 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 12472 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 12473 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.188, + "step": 12474 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1576, + "step": 12475 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1611, + "step": 12476 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.2063, + "step": 12477 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 12478 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1595, + "step": 12479 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1762, + "step": 12480 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.2262, + "step": 12481 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 12482 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1488, + "step": 12483 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1663, + "step": 12484 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 12485 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1542, + "step": 12486 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1848, + "step": 12487 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 12488 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 12489 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1648, + "step": 12490 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1429, + "step": 12491 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1797, + "step": 12492 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1679, + "step": 12493 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1593, + "step": 12494 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1469, + "step": 12495 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 12496 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 12497 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 12498 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.156, + "step": 12499 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1757, + "step": 12500 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 12501 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1924, + "step": 12502 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.145, + "step": 12503 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 12504 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.2035, + "step": 12505 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 12506 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 12507 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 12508 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.2312, + "step": 12509 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1562, + "step": 12510 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.185, + "step": 12511 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 12512 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.167, + "step": 12513 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1611, + "step": 12514 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1728, + "step": 12515 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 12516 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1826, + "step": 12517 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 12518 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 12519 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 12520 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1996, + "step": 12521 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 12522 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1834, + "step": 12523 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1528, + "step": 12524 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 12525 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1624, + "step": 12526 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 12527 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 12528 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1721, + "step": 12529 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1513, + "step": 12530 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1976, + "step": 12531 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 12532 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.2185, + "step": 12533 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.184, + "step": 12534 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1838, + "step": 12535 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 12536 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 12537 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 12538 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 12539 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 12540 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1762, + "step": 12541 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 12542 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 12543 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1894, + "step": 12544 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 12545 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 12546 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.2256, + "step": 12547 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1669, + "step": 12548 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1714, + "step": 12549 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.2213, + "step": 12550 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1554, + "step": 12551 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 12552 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1622, + "step": 12553 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1721, + "step": 12554 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1561, + "step": 12555 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.164, + "step": 12556 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.2022, + "step": 12557 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 12558 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1991, + "step": 12559 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1714, + "step": 12560 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.2217, + "step": 12561 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 12562 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1987, + "step": 12563 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 12564 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.145, + "step": 12565 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 12566 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 12567 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 12568 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 12569 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 12570 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1648, + "step": 12571 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1588, + "step": 12572 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.177, + "step": 12573 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1733, + "step": 12574 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1588, + "step": 12575 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.153, + "step": 12576 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 12577 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 12578 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 12579 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 12580 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 12581 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1625, + "step": 12582 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.204, + "step": 12583 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1555, + "step": 12584 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1648, + "step": 12585 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1485, + "step": 12586 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 12587 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1622, + "step": 12588 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1628, + "step": 12589 + }, + { + "epoch": 0.76, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 12590 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 12591 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 12592 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.2149, + "step": 12593 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 12594 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1549, + "step": 12595 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.148, + "step": 12596 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 12597 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 12598 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 12599 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1593, + "step": 12600 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1625, + "step": 12601 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1719, + "step": 12602 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 12603 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 12604 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 12605 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1697, + "step": 12606 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 12607 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1767, + "step": 12608 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1513, + "step": 12609 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1937, + "step": 12610 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 12611 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1558, + "step": 12612 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1542, + "step": 12613 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1955, + "step": 12614 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 12615 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1498, + "step": 12616 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1718, + "step": 12617 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.159, + "step": 12618 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.2234, + "step": 12619 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 12620 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1595, + "step": 12621 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1623, + "step": 12622 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 12623 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1697, + "step": 12624 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1611, + "step": 12625 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1726, + "step": 12626 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1861, + "step": 12627 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 12628 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.165, + "step": 12629 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1909, + "step": 12630 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 12631 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1696, + "step": 12632 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 12633 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 12634 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.16, + "step": 12635 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1643, + "step": 12636 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.173, + "step": 12637 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1551, + "step": 12638 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1697, + "step": 12639 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1532, + "step": 12640 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.18, + "step": 12641 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 12642 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 12643 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1927, + "step": 12644 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1721, + "step": 12645 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 12646 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1533, + "step": 12647 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1555, + "step": 12648 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1629, + "step": 12649 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 12650 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.169, + "step": 12651 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 12652 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1686, + "step": 12653 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 12654 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.164, + "step": 12655 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 12656 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 12657 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 12658 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1442, + "step": 12659 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1772, + "step": 12660 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1652, + "step": 12661 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.2037, + "step": 12662 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1528, + "step": 12663 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1531, + "step": 12664 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1641, + "step": 12665 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1601, + "step": 12666 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 12667 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1968, + "step": 12668 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1595, + "step": 12669 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 12670 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1555, + "step": 12671 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 12672 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1564, + "step": 12673 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 12674 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1899, + "step": 12675 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 12676 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 12677 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1566, + "step": 12678 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.2156, + "step": 12679 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 12680 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1684, + "step": 12681 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.2001, + "step": 12682 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 12683 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 12684 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.2419, + "step": 12685 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1599, + "step": 12686 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 12687 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1748, + "step": 12688 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1819, + "step": 12689 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 12690 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1632, + "step": 12691 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 12692 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 12693 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1716, + "step": 12694 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 12695 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.181, + "step": 12696 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1778, + "step": 12697 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 12698 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1715, + "step": 12699 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 12700 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 12701 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 12702 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 12703 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1868, + "step": 12704 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 12705 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 12706 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 12707 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1578, + "step": 12708 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 12709 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 12710 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 12711 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.17, + "step": 12712 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.2026, + "step": 12713 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1901, + "step": 12714 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1772, + "step": 12715 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1825, + "step": 12716 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1647, + "step": 12717 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 12718 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 12719 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1727, + "step": 12720 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1796, + "step": 12721 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 12722 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 12723 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 12724 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1734, + "step": 12725 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1562, + "step": 12726 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.189, + "step": 12727 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.175, + "step": 12728 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1743, + "step": 12729 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 12730 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1794, + "step": 12731 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 12732 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1622, + "step": 12733 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1726, + "step": 12734 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1561, + "step": 12735 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 12736 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 12737 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 12738 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.2275, + "step": 12739 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 12740 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.18, + "step": 12741 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 12742 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 12743 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1908, + "step": 12744 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.154, + "step": 12745 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 12746 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1879, + "step": 12747 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.2129, + "step": 12748 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.2042, + "step": 12749 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1561, + "step": 12750 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.2133, + "step": 12751 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1832, + "step": 12752 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1933, + "step": 12753 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 12754 + }, + { + "epoch": 0.77, + "learning_rate": 2e-05, + "loss": 0.1997, + "step": 12755 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1637, + "step": 12756 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 12757 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1714, + "step": 12758 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 12759 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1476, + "step": 12760 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1786, + "step": 12761 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1589, + "step": 12762 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1858, + "step": 12763 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1593, + "step": 12764 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 12765 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 12766 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 12767 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 12768 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 12769 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 12770 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1613, + "step": 12771 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.213, + "step": 12772 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.2204, + "step": 12773 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.183, + "step": 12774 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.2087, + "step": 12775 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 12776 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 12777 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.189, + "step": 12778 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 12779 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1613, + "step": 12780 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 12781 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1848, + "step": 12782 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1462, + "step": 12783 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.191, + "step": 12784 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 12785 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 12786 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1644, + "step": 12787 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1606, + "step": 12788 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1925, + "step": 12789 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1519, + "step": 12790 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1724, + "step": 12791 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 12792 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 12793 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1767, + "step": 12794 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1766, + "step": 12795 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1595, + "step": 12796 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.2222, + "step": 12797 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1741, + "step": 12798 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.154, + "step": 12799 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1542, + "step": 12800 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1687, + "step": 12801 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1864, + "step": 12802 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 12803 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1964, + "step": 12804 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 12805 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 12806 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 12807 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1652, + "step": 12808 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1734, + "step": 12809 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1831, + "step": 12810 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.2049, + "step": 12811 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1799, + "step": 12812 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1886, + "step": 12813 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1651, + "step": 12814 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1978, + "step": 12815 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1672, + "step": 12816 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1904, + "step": 12817 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.175, + "step": 12818 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1746, + "step": 12819 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 12820 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 12821 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1561, + "step": 12822 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1549, + "step": 12823 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1755, + "step": 12824 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1739, + "step": 12825 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1755, + "step": 12826 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.161, + "step": 12827 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 12828 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 12829 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1795, + "step": 12830 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.158, + "step": 12831 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 12832 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 12833 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1848, + "step": 12834 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 12835 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1888, + "step": 12836 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 12837 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 12838 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1631, + "step": 12839 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 12840 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1614, + "step": 12841 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1805, + "step": 12842 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1728, + "step": 12843 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 12844 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 12845 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.177, + "step": 12846 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 12847 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1701, + "step": 12848 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1671, + "step": 12849 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 12850 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1771, + "step": 12851 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1715, + "step": 12852 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1874, + "step": 12853 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 12854 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 12855 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.146, + "step": 12856 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1442, + "step": 12857 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1631, + "step": 12858 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1821, + "step": 12859 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1614, + "step": 12860 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 12861 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 12862 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.2168, + "step": 12863 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1592, + "step": 12864 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1688, + "step": 12865 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1686, + "step": 12866 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 12867 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 12868 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 12869 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 12870 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1774, + "step": 12871 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1767, + "step": 12872 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1722, + "step": 12873 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1757, + "step": 12874 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1719, + "step": 12875 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1516, + "step": 12876 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 12877 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 12878 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1628, + "step": 12879 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1722, + "step": 12880 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1623, + "step": 12881 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.153, + "step": 12882 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1717, + "step": 12883 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 12884 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1593, + "step": 12885 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1645, + "step": 12886 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 12887 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1557, + "step": 12888 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1676, + "step": 12889 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.163, + "step": 12890 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1761, + "step": 12891 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.2034, + "step": 12892 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1804, + "step": 12893 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1616, + "step": 12894 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1696, + "step": 12895 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.2077, + "step": 12896 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 12897 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 12898 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1664, + "step": 12899 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1997, + "step": 12900 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1553, + "step": 12901 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.183, + "step": 12902 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 12903 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 12904 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 12905 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1582, + "step": 12906 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 12907 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1977, + "step": 12908 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1743, + "step": 12909 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.2301, + "step": 12910 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 12911 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1521, + "step": 12912 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1817, + "step": 12913 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1675, + "step": 12914 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1693, + "step": 12915 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 12916 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1941, + "step": 12917 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 12918 + }, + { + "epoch": 0.78, + "learning_rate": 2e-05, + "loss": 0.1805, + "step": 12919 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1588, + "step": 12920 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1845, + "step": 12921 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.2006, + "step": 12922 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1649, + "step": 12923 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 12924 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 12925 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1716, + "step": 12926 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.2152, + "step": 12927 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 12928 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1661, + "step": 12929 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.2055, + "step": 12930 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 12931 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1595, + "step": 12932 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 12933 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1789, + "step": 12934 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 12935 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1725, + "step": 12936 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 12937 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1576, + "step": 12938 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 12939 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 12940 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1819, + "step": 12941 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 12942 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 12943 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.202, + "step": 12944 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.187, + "step": 12945 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1663, + "step": 12946 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 12947 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.2046, + "step": 12948 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1769, + "step": 12949 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.159, + "step": 12950 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.164, + "step": 12951 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1676, + "step": 12952 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1748, + "step": 12953 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 12954 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1546, + "step": 12955 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.175, + "step": 12956 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 12957 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1671, + "step": 12958 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1915, + "step": 12959 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1515, + "step": 12960 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1837, + "step": 12961 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 12962 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.2005, + "step": 12963 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 12964 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1576, + "step": 12965 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 12966 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1704, + "step": 12967 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 12968 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1699, + "step": 12969 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1534, + "step": 12970 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 12971 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1459, + "step": 12972 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 12973 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 12974 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.16, + "step": 12975 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 12976 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1611, + "step": 12977 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 12978 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 12979 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 12980 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1754, + "step": 12981 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1725, + "step": 12982 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 12983 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 12984 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 12985 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 12986 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 12987 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1627, + "step": 12988 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1705, + "step": 12989 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 12990 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 12991 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 12992 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1704, + "step": 12993 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1563, + "step": 12994 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 12995 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 12996 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.193, + "step": 12997 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 12998 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1443, + "step": 12999 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 13000 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1905, + "step": 13001 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 13002 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1868, + "step": 13003 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 13004 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1822, + "step": 13005 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 13006 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1533, + "step": 13007 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1607, + "step": 13008 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 13009 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1913, + "step": 13010 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 13011 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 13012 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1716, + "step": 13013 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1726, + "step": 13014 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 13015 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1646, + "step": 13016 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 13017 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1667, + "step": 13018 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.2047, + "step": 13019 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 13020 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1659, + "step": 13021 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 13022 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1769, + "step": 13023 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.0858, + "step": 13024 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 13025 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1961, + "step": 13026 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1673, + "step": 13027 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1691, + "step": 13028 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 13029 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 13030 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1668, + "step": 13031 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 13032 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.159, + "step": 13033 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 13034 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1596, + "step": 13035 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 13036 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.2406, + "step": 13037 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.2045, + "step": 13038 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1932, + "step": 13039 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 13040 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1649, + "step": 13041 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 13042 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1696, + "step": 13043 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1853, + "step": 13044 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1488, + "step": 13045 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 13046 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 13047 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1712, + "step": 13048 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 13049 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 13050 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 13051 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1695, + "step": 13052 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 13053 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 13054 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.2323, + "step": 13055 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 13056 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1616, + "step": 13057 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.163, + "step": 13058 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1546, + "step": 13059 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1818, + "step": 13060 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 13061 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1667, + "step": 13062 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 13063 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 13064 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1675, + "step": 13065 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 13066 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1645, + "step": 13067 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.2129, + "step": 13068 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1596, + "step": 13069 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1468, + "step": 13070 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1527, + "step": 13071 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1871, + "step": 13072 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 13073 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 13074 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 13075 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.178, + "step": 13076 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 13077 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1541, + "step": 13078 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 13079 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 13080 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1511, + "step": 13081 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 13082 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1912, + "step": 13083 + }, + { + "epoch": 0.79, + "learning_rate": 2e-05, + "loss": 0.1772, + "step": 13084 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1893, + "step": 13085 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 13086 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 13087 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 13088 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 13089 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 13090 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 13091 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1611, + "step": 13092 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.204, + "step": 13093 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 13094 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 13095 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1589, + "step": 13096 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1968, + "step": 13097 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 13098 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1732, + "step": 13099 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 13100 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1764, + "step": 13101 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 13102 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1865, + "step": 13103 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 13104 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 13105 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 13106 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.183, + "step": 13107 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 13108 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1701, + "step": 13109 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1742, + "step": 13110 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 13111 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1596, + "step": 13112 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.185, + "step": 13113 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 13114 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1743, + "step": 13115 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 13116 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1662, + "step": 13117 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1453, + "step": 13118 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1577, + "step": 13119 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1635, + "step": 13120 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1511, + "step": 13121 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 13122 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1592, + "step": 13123 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1932, + "step": 13124 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 13125 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.2063, + "step": 13126 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1924, + "step": 13127 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1856, + "step": 13128 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 13129 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1774, + "step": 13130 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1469, + "step": 13131 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1971, + "step": 13132 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1595, + "step": 13133 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1918, + "step": 13134 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 13135 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1603, + "step": 13136 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 13137 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.2017, + "step": 13138 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 13139 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 13140 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.162, + "step": 13141 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1628, + "step": 13142 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1915, + "step": 13143 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 13144 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1776, + "step": 13145 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1698, + "step": 13146 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1992, + "step": 13147 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 13148 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1739, + "step": 13149 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 13150 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.193, + "step": 13151 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1879, + "step": 13152 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.2391, + "step": 13153 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1787, + "step": 13154 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.195, + "step": 13155 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.2063, + "step": 13156 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 13157 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 13158 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 13159 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1867, + "step": 13160 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 13161 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 13162 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1685, + "step": 13163 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1765, + "step": 13164 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.2273, + "step": 13165 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 13166 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.154, + "step": 13167 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1702, + "step": 13168 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1579, + "step": 13169 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 13170 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 13171 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1851, + "step": 13172 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1638, + "step": 13173 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1642, + "step": 13174 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 13175 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 13176 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1622, + "step": 13177 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 13178 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 13179 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1837, + "step": 13180 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1899, + "step": 13181 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1639, + "step": 13182 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 13183 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1731, + "step": 13184 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.2052, + "step": 13185 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1507, + "step": 13186 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 13187 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1798, + "step": 13188 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1801, + "step": 13189 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 13190 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 13191 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 13192 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 13193 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1868, + "step": 13194 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1803, + "step": 13195 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1716, + "step": 13196 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1532, + "step": 13197 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1478, + "step": 13198 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 13199 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1702, + "step": 13200 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1535, + "step": 13201 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 13202 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.185, + "step": 13203 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1576, + "step": 13204 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.197, + "step": 13205 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.16, + "step": 13206 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1753, + "step": 13207 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.2271, + "step": 13208 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 13209 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 13210 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 13211 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1515, + "step": 13212 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1596, + "step": 13213 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1987, + "step": 13214 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 13215 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1807, + "step": 13216 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 13217 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.179, + "step": 13218 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1668, + "step": 13219 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1933, + "step": 13220 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1967, + "step": 13221 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 13222 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 13223 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 13224 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1657, + "step": 13225 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.2271, + "step": 13226 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1795, + "step": 13227 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 13228 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 13229 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 13230 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 13231 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 13232 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1771, + "step": 13233 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1694, + "step": 13234 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.144, + "step": 13235 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 13236 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 13237 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 13238 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1949, + "step": 13239 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 13240 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.2107, + "step": 13241 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1891, + "step": 13242 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1785, + "step": 13243 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1586, + "step": 13244 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1802, + "step": 13245 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 13246 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1902, + "step": 13247 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 13248 + }, + { + "epoch": 0.8, + "learning_rate": 2e-05, + "loss": 0.1576, + "step": 13249 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1676, + "step": 13250 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1647, + "step": 13251 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1797, + "step": 13252 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1942, + "step": 13253 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 13254 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.176, + "step": 13255 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1576, + "step": 13256 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.161, + "step": 13257 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1782, + "step": 13258 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.2045, + "step": 13259 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1526, + "step": 13260 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1697, + "step": 13261 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1885, + "step": 13262 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.0929, + "step": 13263 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1641, + "step": 13264 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1697, + "step": 13265 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 13266 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1538, + "step": 13267 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1762, + "step": 13268 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 13269 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 13270 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 13271 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1553, + "step": 13272 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1562, + "step": 13273 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1562, + "step": 13274 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.2096, + "step": 13275 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 13276 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1754, + "step": 13277 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1944, + "step": 13278 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 13279 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 13280 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1451, + "step": 13281 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 13282 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.196, + "step": 13283 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1525, + "step": 13284 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1589, + "step": 13285 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1862, + "step": 13286 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1908, + "step": 13287 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1716, + "step": 13288 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 13289 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1811, + "step": 13290 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1765, + "step": 13291 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.153, + "step": 13292 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1836, + "step": 13293 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.2105, + "step": 13294 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1549, + "step": 13295 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1723, + "step": 13296 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1799, + "step": 13297 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1676, + "step": 13298 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 13299 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1939, + "step": 13300 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 13301 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.2128, + "step": 13302 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1712, + "step": 13303 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1648, + "step": 13304 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1497, + "step": 13305 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1774, + "step": 13306 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1877, + "step": 13307 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1808, + "step": 13308 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1784, + "step": 13309 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 13310 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.185, + "step": 13311 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 13312 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 13313 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 13314 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1826, + "step": 13315 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1469, + "step": 13316 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1642, + "step": 13317 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1652, + "step": 13318 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 13319 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1781, + "step": 13320 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1866, + "step": 13321 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 13322 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1447, + "step": 13323 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 13324 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1739, + "step": 13325 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1476, + "step": 13326 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 13327 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1847, + "step": 13328 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1552, + "step": 13329 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.163, + "step": 13330 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 13331 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1804, + "step": 13332 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 13333 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 13334 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 13335 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 13336 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 13337 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 13338 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 13339 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 13340 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1767, + "step": 13341 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 13342 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 13343 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 13344 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 13345 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1526, + "step": 13346 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1667, + "step": 13347 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1812, + "step": 13348 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1757, + "step": 13349 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 13350 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 13351 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1746, + "step": 13352 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1899, + "step": 13353 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1998, + "step": 13354 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1941, + "step": 13355 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.2026, + "step": 13356 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 13357 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.2018, + "step": 13358 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 13359 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1591, + "step": 13360 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.2552, + "step": 13361 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 13362 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 13363 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 13364 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1961, + "step": 13365 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 13366 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.2055, + "step": 13367 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1885, + "step": 13368 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 13369 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 13370 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1459, + "step": 13371 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 13372 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1726, + "step": 13373 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.154, + "step": 13374 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 13375 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 13376 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1919, + "step": 13377 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1761, + "step": 13378 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1894, + "step": 13379 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1703, + "step": 13380 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1766, + "step": 13381 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1797, + "step": 13382 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 13383 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1696, + "step": 13384 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.2038, + "step": 13385 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 13386 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 13387 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 13388 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 13389 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 13390 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 13391 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1872, + "step": 13392 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.2158, + "step": 13393 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1888, + "step": 13394 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1805, + "step": 13395 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 13396 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.184, + "step": 13397 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 13398 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1502, + "step": 13399 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 13400 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1695, + "step": 13401 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 13402 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1667, + "step": 13403 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 13404 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1795, + "step": 13405 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 13406 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1647, + "step": 13407 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1769, + "step": 13408 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 13409 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.2245, + "step": 13410 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1721, + "step": 13411 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.1832, + "step": 13412 + }, + { + "epoch": 0.81, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 13413 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1894, + "step": 13414 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1796, + "step": 13415 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1687, + "step": 13416 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 13417 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 13418 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.2223, + "step": 13419 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1931, + "step": 13420 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1905, + "step": 13421 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1776, + "step": 13422 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 13423 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1513, + "step": 13424 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 13425 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.2006, + "step": 13426 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 13427 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.2013, + "step": 13428 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.181, + "step": 13429 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1899, + "step": 13430 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1733, + "step": 13431 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 13432 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 13433 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 13434 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1859, + "step": 13435 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1714, + "step": 13436 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 13437 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1933, + "step": 13438 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 13439 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1836, + "step": 13440 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1948, + "step": 13441 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1579, + "step": 13442 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.227, + "step": 13443 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1596, + "step": 13444 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 13445 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1658, + "step": 13446 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 13447 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 13448 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 13449 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1849, + "step": 13450 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.196, + "step": 13451 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 13452 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1745, + "step": 13453 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 13454 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 13455 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.2242, + "step": 13456 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 13457 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1827, + "step": 13458 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 13459 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1456, + "step": 13460 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.144, + "step": 13461 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1764, + "step": 13462 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1806, + "step": 13463 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.2209, + "step": 13464 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 13465 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 13466 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1711, + "step": 13467 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1879, + "step": 13468 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.22, + "step": 13469 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.2109, + "step": 13470 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 13471 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1779, + "step": 13472 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1723, + "step": 13473 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1747, + "step": 13474 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 13475 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 13476 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.178, + "step": 13477 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1993, + "step": 13478 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1827, + "step": 13479 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 13480 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 13481 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 13482 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1979, + "step": 13483 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 13484 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 13485 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 13486 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1843, + "step": 13487 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 13488 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 13489 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1727, + "step": 13490 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1608, + "step": 13491 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1763, + "step": 13492 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 13493 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.2067, + "step": 13494 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1621, + "step": 13495 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 13496 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 13497 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1738, + "step": 13498 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1691, + "step": 13499 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1752, + "step": 13500 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1722, + "step": 13501 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1595, + "step": 13502 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1667, + "step": 13503 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1478, + "step": 13504 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 13505 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.176, + "step": 13506 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1788, + "step": 13507 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1793, + "step": 13508 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 13509 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.171, + "step": 13510 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1652, + "step": 13511 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 13512 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 13513 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1617, + "step": 13514 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.2328, + "step": 13515 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 13516 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 13517 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1456, + "step": 13518 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1745, + "step": 13519 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 13520 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 13521 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1958, + "step": 13522 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.186, + "step": 13523 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 13524 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1978, + "step": 13525 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 13526 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.176, + "step": 13527 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1717, + "step": 13528 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 13529 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 13530 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1601, + "step": 13531 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 13532 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1928, + "step": 13533 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1825, + "step": 13534 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1703, + "step": 13535 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1733, + "step": 13536 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 13537 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.2616, + "step": 13538 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 13539 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1711, + "step": 13540 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 13541 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 13542 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 13543 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 13544 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.2297, + "step": 13545 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1553, + "step": 13546 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1737, + "step": 13547 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 13548 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1779, + "step": 13549 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 13550 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 13551 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.2099, + "step": 13552 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 13553 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1691, + "step": 13554 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1696, + "step": 13555 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1815, + "step": 13556 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1776, + "step": 13557 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1704, + "step": 13558 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1902, + "step": 13559 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.198, + "step": 13560 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1741, + "step": 13561 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.2075, + "step": 13562 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1771, + "step": 13563 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 13564 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1816, + "step": 13565 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 13566 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1677, + "step": 13567 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 13568 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 13569 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 13570 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1979, + "step": 13571 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.2133, + "step": 13572 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 13573 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 13574 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1601, + "step": 13575 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1613, + "step": 13576 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1939, + "step": 13577 + }, + { + "epoch": 0.82, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 13578 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1728, + "step": 13579 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1817, + "step": 13580 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 13581 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1932, + "step": 13582 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1697, + "step": 13583 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 13584 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1868, + "step": 13585 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1589, + "step": 13586 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1684, + "step": 13587 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.2093, + "step": 13588 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 13589 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.175, + "step": 13590 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1937, + "step": 13591 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1981, + "step": 13592 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 13593 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 13594 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 13595 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.167, + "step": 13596 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.181, + "step": 13597 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 13598 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1606, + "step": 13599 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 13600 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 13601 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1497, + "step": 13602 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1962, + "step": 13603 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1951, + "step": 13604 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1541, + "step": 13605 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1737, + "step": 13606 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.2305, + "step": 13607 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1846, + "step": 13608 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 13609 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1549, + "step": 13610 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1521, + "step": 13611 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 13612 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 13613 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1633, + "step": 13614 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1706, + "step": 13615 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1748, + "step": 13616 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 13617 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 13618 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 13619 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 13620 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1844, + "step": 13621 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1683, + "step": 13622 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 13623 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 13624 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.2011, + "step": 13625 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 13626 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1562, + "step": 13627 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 13628 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1599, + "step": 13629 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 13630 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1662, + "step": 13631 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 13632 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.2003, + "step": 13633 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1557, + "step": 13634 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1644, + "step": 13635 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 13636 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 13637 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1655, + "step": 13638 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1746, + "step": 13639 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.2148, + "step": 13640 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1739, + "step": 13641 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 13642 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1787, + "step": 13643 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1848, + "step": 13644 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1645, + "step": 13645 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1762, + "step": 13646 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 13647 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 13648 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1652, + "step": 13649 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.162, + "step": 13650 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1864, + "step": 13651 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1521, + "step": 13652 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1884, + "step": 13653 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 13654 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 13655 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1713, + "step": 13656 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1715, + "step": 13657 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1718, + "step": 13658 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 13659 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 13660 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1553, + "step": 13661 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1518, + "step": 13662 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 13663 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1691, + "step": 13664 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 13665 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 13666 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 13667 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1478, + "step": 13668 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1459, + "step": 13669 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1981, + "step": 13670 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1524, + "step": 13671 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1593, + "step": 13672 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 13673 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.156, + "step": 13674 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 13675 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1478, + "step": 13676 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1952, + "step": 13677 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1531, + "step": 13678 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 13679 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1847, + "step": 13680 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1535, + "step": 13681 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 13682 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1852, + "step": 13683 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.177, + "step": 13684 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 13685 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 13686 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1696, + "step": 13687 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.165, + "step": 13688 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 13689 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1697, + "step": 13690 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 13691 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1879, + "step": 13692 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 13693 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1517, + "step": 13694 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1933, + "step": 13695 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 13696 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.208, + "step": 13697 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1819, + "step": 13698 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1954, + "step": 13699 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1784, + "step": 13700 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1662, + "step": 13701 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1793, + "step": 13702 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 13703 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1743, + "step": 13704 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 13705 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 13706 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1711, + "step": 13707 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 13708 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 13709 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 13710 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 13711 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 13712 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1665, + "step": 13713 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1875, + "step": 13714 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 13715 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 13716 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1871, + "step": 13717 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1631, + "step": 13718 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 13719 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1456, + "step": 13720 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 13721 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 13722 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 13723 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1606, + "step": 13724 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1533, + "step": 13725 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 13726 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 13727 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1673, + "step": 13728 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 13729 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1564, + "step": 13730 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 13731 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 13732 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 13733 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.2075, + "step": 13734 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 13735 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1689, + "step": 13736 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1635, + "step": 13737 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.146, + "step": 13738 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1488, + "step": 13739 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1813, + "step": 13740 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.172, + "step": 13741 + }, + { + "epoch": 0.83, + "learning_rate": 2e-05, + "loss": 0.1824, + "step": 13742 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1723, + "step": 13743 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1488, + "step": 13744 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 13745 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 13746 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1902, + "step": 13747 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1844, + "step": 13748 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1969, + "step": 13749 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 13750 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1687, + "step": 13751 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1909, + "step": 13752 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 13753 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1801, + "step": 13754 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 13755 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 13756 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1878, + "step": 13757 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 13758 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1756, + "step": 13759 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1706, + "step": 13760 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 13761 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1711, + "step": 13762 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1638, + "step": 13763 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1608, + "step": 13764 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.2118, + "step": 13765 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.154, + "step": 13766 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 13767 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1747, + "step": 13768 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1735, + "step": 13769 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 13770 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1939, + "step": 13771 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1646, + "step": 13772 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1971, + "step": 13773 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 13774 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.2222, + "step": 13775 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 13776 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1608, + "step": 13777 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1925, + "step": 13778 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1809, + "step": 13779 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 13780 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1563, + "step": 13781 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 13782 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.174, + "step": 13783 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 13784 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.179, + "step": 13785 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1653, + "step": 13786 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1977, + "step": 13787 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.156, + "step": 13788 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 13789 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1527, + "step": 13790 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 13791 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 13792 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1608, + "step": 13793 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1573, + "step": 13794 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 13795 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1634, + "step": 13796 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.195, + "step": 13797 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 13798 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1719, + "step": 13799 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1632, + "step": 13800 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 13801 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 13802 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 13803 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 13804 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1645, + "step": 13805 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1954, + "step": 13806 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1751, + "step": 13807 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 13808 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 13809 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1573, + "step": 13810 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 13811 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 13812 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 13813 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.159, + "step": 13814 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 13815 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.161, + "step": 13816 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 13817 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1589, + "step": 13818 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 13819 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1877, + "step": 13820 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1517, + "step": 13821 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1812, + "step": 13822 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1442, + "step": 13823 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 13824 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1804, + "step": 13825 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 13826 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 13827 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1876, + "step": 13828 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 13829 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1759, + "step": 13830 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 13831 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.2001, + "step": 13832 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1767, + "step": 13833 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1766, + "step": 13834 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.2393, + "step": 13835 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 13836 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 13837 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1964, + "step": 13838 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 13839 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.201, + "step": 13840 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 13841 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 13842 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1532, + "step": 13843 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.145, + "step": 13844 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1821, + "step": 13845 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1625, + "step": 13846 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1852, + "step": 13847 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1632, + "step": 13848 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.2091, + "step": 13849 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1954, + "step": 13850 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1748, + "step": 13851 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1546, + "step": 13852 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1654, + "step": 13853 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1728, + "step": 13854 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.2216, + "step": 13855 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1745, + "step": 13856 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 13857 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.182, + "step": 13858 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1621, + "step": 13859 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 13860 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1669, + "step": 13861 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 13862 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1965, + "step": 13863 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.152, + "step": 13864 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1593, + "step": 13865 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.2006, + "step": 13866 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.2006, + "step": 13867 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1923, + "step": 13868 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 13869 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 13870 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1663, + "step": 13871 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.2008, + "step": 13872 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1948, + "step": 13873 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 13874 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 13875 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1603, + "step": 13876 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 13877 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 13878 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 13879 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1447, + "step": 13880 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 13881 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 13882 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1691, + "step": 13883 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 13884 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.2018, + "step": 13885 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1618, + "step": 13886 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1796, + "step": 13887 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 13888 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.185, + "step": 13889 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1757, + "step": 13890 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1628, + "step": 13891 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1769, + "step": 13892 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 13893 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 13894 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1429, + "step": 13895 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1604, + "step": 13896 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1878, + "step": 13897 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1684, + "step": 13898 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 13899 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.2002, + "step": 13900 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1827, + "step": 13901 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 13902 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 13903 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1953, + "step": 13904 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 13905 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 13906 + }, + { + "epoch": 0.84, + "learning_rate": 2e-05, + "loss": 0.1718, + "step": 13907 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1753, + "step": 13908 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.188, + "step": 13909 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 13910 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 13911 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 13912 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 13913 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 13914 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 13915 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 13916 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 13917 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1598, + "step": 13918 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1617, + "step": 13919 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 13920 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 13921 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1672, + "step": 13922 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 13923 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.2229, + "step": 13924 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.2073, + "step": 13925 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 13926 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 13927 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1672, + "step": 13928 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1847, + "step": 13929 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 13930 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.2027, + "step": 13931 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 13932 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1715, + "step": 13933 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1835, + "step": 13934 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 13935 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1532, + "step": 13936 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1777, + "step": 13937 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1653, + "step": 13938 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 13939 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 13940 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 13941 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 13942 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 13943 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.0967, + "step": 13944 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1563, + "step": 13945 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1926, + "step": 13946 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 13947 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.2025, + "step": 13948 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 13949 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1966, + "step": 13950 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1602, + "step": 13951 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1462, + "step": 13952 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 13953 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 13954 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.21, + "step": 13955 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1809, + "step": 13956 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1531, + "step": 13957 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1723, + "step": 13958 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 13959 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1476, + "step": 13960 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1659, + "step": 13961 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 13962 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1958, + "step": 13963 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1588, + "step": 13964 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 13965 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1862, + "step": 13966 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 13967 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1631, + "step": 13968 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1885, + "step": 13969 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1732, + "step": 13970 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.174, + "step": 13971 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1803, + "step": 13972 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 13973 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1906, + "step": 13974 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1869, + "step": 13975 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.18, + "step": 13976 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 13977 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1579, + "step": 13978 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 13979 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 13980 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1517, + "step": 13981 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 13982 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1542, + "step": 13983 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 13984 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 13985 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.145, + "step": 13986 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1499, + "step": 13987 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1704, + "step": 13988 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 13989 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1781, + "step": 13990 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1773, + "step": 13991 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 13992 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.171, + "step": 13993 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1476, + "step": 13994 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.2156, + "step": 13995 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 13996 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 13997 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 13998 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 13999 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1646, + "step": 14000 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1548, + "step": 14001 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 14002 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.2191, + "step": 14003 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 14004 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.181, + "step": 14005 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 14006 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 14007 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1588, + "step": 14008 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1649, + "step": 14009 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1447, + "step": 14010 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 14011 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 14012 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 14013 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 14014 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 14015 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 14016 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.2066, + "step": 14017 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1451, + "step": 14018 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1946, + "step": 14019 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 14020 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 14021 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1712, + "step": 14022 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1532, + "step": 14023 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1719, + "step": 14024 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.2193, + "step": 14025 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1736, + "step": 14026 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1762, + "step": 14027 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 14028 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1459, + "step": 14029 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 14030 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.175, + "step": 14031 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1725, + "step": 14032 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.167, + "step": 14033 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1653, + "step": 14034 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.213, + "step": 14035 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 14036 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1854, + "step": 14037 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1923, + "step": 14038 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 14039 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 14040 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1791, + "step": 14041 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1626, + "step": 14042 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1599, + "step": 14043 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1958, + "step": 14044 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1834, + "step": 14045 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1902, + "step": 14046 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1821, + "step": 14047 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 14048 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.2025, + "step": 14049 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1607, + "step": 14050 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 14051 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 14052 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1902, + "step": 14053 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 14054 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.2051, + "step": 14055 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1682, + "step": 14056 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1796, + "step": 14057 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1829, + "step": 14058 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 14059 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1985, + "step": 14060 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.2169, + "step": 14061 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 14062 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 14063 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 14064 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 14065 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 14066 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 14067 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 14068 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 14069 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 14070 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.1843, + "step": 14071 + }, + { + "epoch": 0.85, + "learning_rate": 2e-05, + "loss": 0.165, + "step": 14072 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1766, + "step": 14073 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1582, + "step": 14074 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.203, + "step": 14075 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 14076 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1684, + "step": 14077 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 14078 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 14079 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 14080 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1562, + "step": 14081 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1757, + "step": 14082 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1777, + "step": 14083 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 14084 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1914, + "step": 14085 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1527, + "step": 14086 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1623, + "step": 14087 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1923, + "step": 14088 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1998, + "step": 14089 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1542, + "step": 14090 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1898, + "step": 14091 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1961, + "step": 14092 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1563, + "step": 14093 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 14094 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1753, + "step": 14095 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1966, + "step": 14096 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.2031, + "step": 14097 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 14098 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 14099 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1848, + "step": 14100 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 14101 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1707, + "step": 14102 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 14103 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1485, + "step": 14104 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1784, + "step": 14105 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1715, + "step": 14106 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1996, + "step": 14107 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 14108 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1785, + "step": 14109 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 14110 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 14111 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1579, + "step": 14112 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 14113 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1562, + "step": 14114 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1793, + "step": 14115 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 14116 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 14117 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 14118 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1814, + "step": 14119 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.2019, + "step": 14120 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 14121 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 14122 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 14123 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1888, + "step": 14124 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1649, + "step": 14125 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 14126 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 14127 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 14128 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 14129 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 14130 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1829, + "step": 14131 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1631, + "step": 14132 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 14133 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 14134 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 14135 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 14136 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 14137 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1624, + "step": 14138 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 14139 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 14140 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1605, + "step": 14141 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 14142 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1651, + "step": 14143 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 14144 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 14145 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 14146 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 14147 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 14148 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 14149 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 14150 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 14151 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 14152 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.17, + "step": 14153 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1788, + "step": 14154 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.152, + "step": 14155 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 14156 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.2189, + "step": 14157 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1553, + "step": 14158 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 14159 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1679, + "step": 14160 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 14161 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1706, + "step": 14162 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1798, + "step": 14163 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.224, + "step": 14164 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1607, + "step": 14165 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 14166 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1643, + "step": 14167 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 14168 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1774, + "step": 14169 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 14170 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1519, + "step": 14171 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1699, + "step": 14172 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1592, + "step": 14173 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1815, + "step": 14174 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 14175 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1533, + "step": 14176 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 14177 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 14178 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 14179 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1607, + "step": 14180 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1532, + "step": 14181 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 14182 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1789, + "step": 14183 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 14184 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 14185 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 14186 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1653, + "step": 14187 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1729, + "step": 14188 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1502, + "step": 14189 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1713, + "step": 14190 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1803, + "step": 14191 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1814, + "step": 14192 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1568, + "step": 14193 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 14194 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.192, + "step": 14195 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 14196 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 14197 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1817, + "step": 14198 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1648, + "step": 14199 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1532, + "step": 14200 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1618, + "step": 14201 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1518, + "step": 14202 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 14203 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1505, + "step": 14204 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1914, + "step": 14205 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1627, + "step": 14206 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1672, + "step": 14207 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1631, + "step": 14208 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 14209 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 14210 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1998, + "step": 14211 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1769, + "step": 14212 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1643, + "step": 14213 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1846, + "step": 14214 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 14215 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 14216 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 14217 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.186, + "step": 14218 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1652, + "step": 14219 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1548, + "step": 14220 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 14221 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 14222 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1736, + "step": 14223 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 14224 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 14225 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1613, + "step": 14226 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 14227 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 14228 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1722, + "step": 14229 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1485, + "step": 14230 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 14231 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 14232 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1911, + "step": 14233 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.185, + "step": 14234 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 14235 + }, + { + "epoch": 0.86, + "learning_rate": 2e-05, + "loss": 0.1878, + "step": 14236 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 14237 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1706, + "step": 14238 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1658, + "step": 14239 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1641, + "step": 14240 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1711, + "step": 14241 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1657, + "step": 14242 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1591, + "step": 14243 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 14244 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1578, + "step": 14245 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1838, + "step": 14246 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 14247 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1716, + "step": 14248 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1488, + "step": 14249 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 14250 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1834, + "step": 14251 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1653, + "step": 14252 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 14253 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1597, + "step": 14254 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 14255 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1648, + "step": 14256 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1533, + "step": 14257 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 14258 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 14259 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.189, + "step": 14260 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1767, + "step": 14261 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 14262 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.154, + "step": 14263 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1707, + "step": 14264 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1769, + "step": 14265 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 14266 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 14267 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1531, + "step": 14268 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1546, + "step": 14269 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1783, + "step": 14270 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.173, + "step": 14271 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.171, + "step": 14272 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1725, + "step": 14273 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.176, + "step": 14274 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 14275 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 14276 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.175, + "step": 14277 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 14278 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1606, + "step": 14279 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 14280 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 14281 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1764, + "step": 14282 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1802, + "step": 14283 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 14284 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.181, + "step": 14285 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1614, + "step": 14286 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 14287 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1566, + "step": 14288 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 14289 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1693, + "step": 14290 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 14291 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1811, + "step": 14292 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1838, + "step": 14293 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1888, + "step": 14294 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1855, + "step": 14295 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1773, + "step": 14296 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1981, + "step": 14297 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1605, + "step": 14298 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1495, + "step": 14299 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 14300 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1873, + "step": 14301 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1774, + "step": 14302 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1953, + "step": 14303 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.189, + "step": 14304 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1546, + "step": 14305 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1578, + "step": 14306 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 14307 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1778, + "step": 14308 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.18, + "step": 14309 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 14310 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 14311 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 14312 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1797, + "step": 14313 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 14314 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.173, + "step": 14315 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 14316 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1624, + "step": 14317 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1661, + "step": 14318 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.2222, + "step": 14319 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.2284, + "step": 14320 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.2565, + "step": 14321 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 14322 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1993, + "step": 14323 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 14324 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.159, + "step": 14325 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 14326 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1883, + "step": 14327 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 14328 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 14329 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.2438, + "step": 14330 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1517, + "step": 14331 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 14332 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 14333 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 14334 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1553, + "step": 14335 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1825, + "step": 14336 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1814, + "step": 14337 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.2545, + "step": 14338 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1766, + "step": 14339 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 14340 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 14341 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1649, + "step": 14342 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 14343 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1688, + "step": 14344 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1626, + "step": 14345 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 14346 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 14347 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1521, + "step": 14348 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1707, + "step": 14349 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1784, + "step": 14350 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1877, + "step": 14351 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1679, + "step": 14352 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 14353 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 14354 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 14355 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 14356 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1634, + "step": 14357 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1703, + "step": 14358 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 14359 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 14360 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1642, + "step": 14361 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1641, + "step": 14362 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 14363 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1534, + "step": 14364 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 14365 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 14366 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 14367 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 14368 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1711, + "step": 14369 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 14370 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.173, + "step": 14371 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.191, + "step": 14372 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1564, + "step": 14373 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1525, + "step": 14374 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 14375 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.174, + "step": 14376 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.146, + "step": 14377 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 14378 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 14379 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1825, + "step": 14380 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1785, + "step": 14381 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1593, + "step": 14382 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 14383 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.165, + "step": 14384 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1667, + "step": 14385 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 14386 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1948, + "step": 14387 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1774, + "step": 14388 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1894, + "step": 14389 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.2171, + "step": 14390 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1624, + "step": 14391 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 14392 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1665, + "step": 14393 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 14394 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.188, + "step": 14395 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.156, + "step": 14396 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 14397 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 14398 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1554, + "step": 14399 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 14400 + }, + { + "epoch": 0.87, + "learning_rate": 2e-05, + "loss": 0.1443, + "step": 14401 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 14402 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1806, + "step": 14403 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1824, + "step": 14404 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1836, + "step": 14405 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1625, + "step": 14406 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1675, + "step": 14407 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.156, + "step": 14408 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 14409 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1478, + "step": 14410 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1551, + "step": 14411 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1657, + "step": 14412 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 14413 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1639, + "step": 14414 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 14415 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 14416 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1595, + "step": 14417 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 14418 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 14419 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 14420 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 14421 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 14422 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 14423 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.208, + "step": 14424 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1656, + "step": 14425 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 14426 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1521, + "step": 14427 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 14428 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 14429 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 14430 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 14431 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 14432 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 14433 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1896, + "step": 14434 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 14435 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1612, + "step": 14436 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.2023, + "step": 14437 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1829, + "step": 14438 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1898, + "step": 14439 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1503, + "step": 14440 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1443, + "step": 14441 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 14442 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.204, + "step": 14443 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.153, + "step": 14444 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1646, + "step": 14445 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.183, + "step": 14446 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1851, + "step": 14447 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 14448 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1747, + "step": 14449 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 14450 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1906, + "step": 14451 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 14452 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.181, + "step": 14453 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.201, + "step": 14454 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1656, + "step": 14455 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1666, + "step": 14456 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.208, + "step": 14457 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1592, + "step": 14458 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.2029, + "step": 14459 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 14460 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1777, + "step": 14461 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.2006, + "step": 14462 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1911, + "step": 14463 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1602, + "step": 14464 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1785, + "step": 14465 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 14466 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1623, + "step": 14467 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 14468 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1628, + "step": 14469 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1696, + "step": 14470 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1527, + "step": 14471 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 14472 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.163, + "step": 14473 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1691, + "step": 14474 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1986, + "step": 14475 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 14476 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 14477 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.2064, + "step": 14478 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.201, + "step": 14479 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1809, + "step": 14480 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 14481 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1526, + "step": 14482 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 14483 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 14484 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1893, + "step": 14485 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1513, + "step": 14486 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1769, + "step": 14487 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1844, + "step": 14488 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1685, + "step": 14489 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 14490 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1682, + "step": 14491 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1692, + "step": 14492 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 14493 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 14494 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.176, + "step": 14495 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 14496 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1954, + "step": 14497 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.2057, + "step": 14498 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 14499 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 14500 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 14501 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1786, + "step": 14502 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1852, + "step": 14503 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 14504 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 14505 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1613, + "step": 14506 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1866, + "step": 14507 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1812, + "step": 14508 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 14509 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.19, + "step": 14510 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 14511 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1663, + "step": 14512 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 14513 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 14514 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1771, + "step": 14515 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1725, + "step": 14516 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1949, + "step": 14517 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1734, + "step": 14518 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1777, + "step": 14519 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 14520 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 14521 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.162, + "step": 14522 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1803, + "step": 14523 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1978, + "step": 14524 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 14525 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 14526 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1582, + "step": 14527 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1903, + "step": 14528 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 14529 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 14530 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1429, + "step": 14531 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 14532 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 14533 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 14534 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 14535 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.208, + "step": 14536 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 14537 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 14538 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 14539 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1875, + "step": 14540 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.16, + "step": 14541 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1827, + "step": 14542 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 14543 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 14544 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 14545 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1844, + "step": 14546 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1502, + "step": 14547 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 14548 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1871, + "step": 14549 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1713, + "step": 14550 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1759, + "step": 14551 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 14552 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 14553 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1688, + "step": 14554 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1647, + "step": 14555 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1833, + "step": 14556 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 14557 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 14558 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.2042, + "step": 14559 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 14560 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 14561 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 14562 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1612, + "step": 14563 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 14564 + }, + { + "epoch": 0.88, + "learning_rate": 2e-05, + "loss": 0.188, + "step": 14565 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1597, + "step": 14566 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1592, + "step": 14567 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 14568 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1525, + "step": 14569 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 14570 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1498, + "step": 14571 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 14572 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1791, + "step": 14573 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1765, + "step": 14574 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 14575 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 14576 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 14577 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 14578 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1787, + "step": 14579 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 14580 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 14581 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1825, + "step": 14582 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 14583 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 14584 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1887, + "step": 14585 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 14586 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1718, + "step": 14587 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 14588 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.153, + "step": 14589 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1942, + "step": 14590 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1528, + "step": 14591 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 14592 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 14593 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 14594 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 14595 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1816, + "step": 14596 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 14597 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1956, + "step": 14598 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.2381, + "step": 14599 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 14600 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1756, + "step": 14601 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1721, + "step": 14602 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 14603 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.208, + "step": 14604 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1598, + "step": 14605 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 14606 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1663, + "step": 14607 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1727, + "step": 14608 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.2012, + "step": 14609 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 14610 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1954, + "step": 14611 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1812, + "step": 14612 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1812, + "step": 14613 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 14614 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1866, + "step": 14615 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1926, + "step": 14616 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1532, + "step": 14617 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1593, + "step": 14618 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 14619 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1888, + "step": 14620 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1583, + "step": 14621 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 14622 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 14623 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.152, + "step": 14624 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1879, + "step": 14625 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 14626 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1904, + "step": 14627 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.211, + "step": 14628 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 14629 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 14630 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1976, + "step": 14631 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1681, + "step": 14632 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 14633 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 14634 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1886, + "step": 14635 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1488, + "step": 14636 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1519, + "step": 14637 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 14638 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1796, + "step": 14639 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1646, + "step": 14640 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1459, + "step": 14641 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1861, + "step": 14642 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1812, + "step": 14643 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 14644 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 14645 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.197, + "step": 14646 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 14647 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1855, + "step": 14648 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1974, + "step": 14649 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.153, + "step": 14650 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 14651 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 14652 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 14653 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 14654 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1734, + "step": 14655 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.2012, + "step": 14656 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.2158, + "step": 14657 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1701, + "step": 14658 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 14659 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 14660 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 14661 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1639, + "step": 14662 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.2176, + "step": 14663 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 14664 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 14665 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 14666 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 14667 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 14668 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 14669 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.2114, + "step": 14670 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1527, + "step": 14671 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1727, + "step": 14672 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.2312, + "step": 14673 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.206, + "step": 14674 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 14675 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 14676 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 14677 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 14678 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 14679 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1659, + "step": 14680 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 14681 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1447, + "step": 14682 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 14683 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 14684 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 14685 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 14686 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1524, + "step": 14687 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1879, + "step": 14688 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 14689 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 14690 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 14691 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 14692 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1742, + "step": 14693 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1502, + "step": 14694 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1966, + "step": 14695 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 14696 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 14697 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 14698 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 14699 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1823, + "step": 14700 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 14701 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1799, + "step": 14702 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1947, + "step": 14703 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 14704 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1564, + "step": 14705 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1551, + "step": 14706 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1651, + "step": 14707 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 14708 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 14709 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 14710 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1634, + "step": 14711 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 14712 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 14713 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1612, + "step": 14714 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1462, + "step": 14715 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 14716 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 14717 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1806, + "step": 14718 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.167, + "step": 14719 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 14720 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1809, + "step": 14721 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1679, + "step": 14722 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 14723 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 14724 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.202, + "step": 14725 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1649, + "step": 14726 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 14727 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1813, + "step": 14728 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1451, + "step": 14729 + }, + { + "epoch": 0.89, + "learning_rate": 2e-05, + "loss": 0.1557, + "step": 14730 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1546, + "step": 14731 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1837, + "step": 14732 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1552, + "step": 14733 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.2014, + "step": 14734 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.156, + "step": 14735 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1799, + "step": 14736 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 14737 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.207, + "step": 14738 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 14739 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1488, + "step": 14740 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1882, + "step": 14741 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1973, + "step": 14742 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 14743 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1842, + "step": 14744 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.152, + "step": 14745 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.2028, + "step": 14746 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1878, + "step": 14747 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 14748 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1507, + "step": 14749 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 14750 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1939, + "step": 14751 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1807, + "step": 14752 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 14753 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.158, + "step": 14754 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1867, + "step": 14755 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1686, + "step": 14756 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.2008, + "step": 14757 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 14758 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1589, + "step": 14759 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1707, + "step": 14760 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1498, + "step": 14761 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1755, + "step": 14762 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 14763 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.153, + "step": 14764 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1778, + "step": 14765 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 14766 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1987, + "step": 14767 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 14768 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1675, + "step": 14769 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1755, + "step": 14770 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 14771 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.146, + "step": 14772 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 14773 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 14774 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1688, + "step": 14775 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 14776 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1674, + "step": 14777 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.154, + "step": 14778 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 14779 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1766, + "step": 14780 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1519, + "step": 14781 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1606, + "step": 14782 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1871, + "step": 14783 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1788, + "step": 14784 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 14785 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 14786 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1875, + "step": 14787 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 14788 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.2057, + "step": 14789 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 14790 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1635, + "step": 14791 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1945, + "step": 14792 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 14793 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1811, + "step": 14794 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 14795 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1653, + "step": 14796 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 14797 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1525, + "step": 14798 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 14799 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1856, + "step": 14800 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.2148, + "step": 14801 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1612, + "step": 14802 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1516, + "step": 14803 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 14804 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1641, + "step": 14805 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1825, + "step": 14806 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 14807 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 14808 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1478, + "step": 14809 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1632, + "step": 14810 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 14811 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1718, + "step": 14812 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.2123, + "step": 14813 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 14814 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 14815 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1707, + "step": 14816 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 14817 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1825, + "step": 14818 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 14819 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 14820 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1738, + "step": 14821 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 14822 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1693, + "step": 14823 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 14824 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 14825 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1749, + "step": 14826 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 14827 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.145, + "step": 14828 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 14829 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1761, + "step": 14830 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 14831 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1627, + "step": 14832 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 14833 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 14834 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1788, + "step": 14835 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 14836 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 14837 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1733, + "step": 14838 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1976, + "step": 14839 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 14840 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 14841 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1743, + "step": 14842 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 14843 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1699, + "step": 14844 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1945, + "step": 14845 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 14846 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 14847 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1966, + "step": 14848 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 14849 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1607, + "step": 14850 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 14851 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 14852 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 14853 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1909, + "step": 14854 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1709, + "step": 14855 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1612, + "step": 14856 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1738, + "step": 14857 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1876, + "step": 14858 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 14859 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1855, + "step": 14860 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.161, + "step": 14861 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1666, + "step": 14862 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1429, + "step": 14863 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1607, + "step": 14864 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 14865 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 14866 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1678, + "step": 14867 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 14868 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 14869 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1554, + "step": 14870 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1517, + "step": 14871 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1772, + "step": 14872 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1553, + "step": 14873 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.165, + "step": 14874 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 14875 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1696, + "step": 14876 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 14877 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 14878 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1675, + "step": 14879 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1655, + "step": 14880 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 14881 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1721, + "step": 14882 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1801, + "step": 14883 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1694, + "step": 14884 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1553, + "step": 14885 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.175, + "step": 14886 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1676, + "step": 14887 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 14888 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 14889 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 14890 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1746, + "step": 14891 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.174, + "step": 14892 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 14893 + }, + { + "epoch": 0.9, + "learning_rate": 2e-05, + "loss": 0.2002, + "step": 14894 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 14895 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 14896 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1948, + "step": 14897 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1548, + "step": 14898 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 14899 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 14900 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1824, + "step": 14901 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 14902 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 14903 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 14904 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 14905 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 14906 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1592, + "step": 14907 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 14908 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 14909 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 14910 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1502, + "step": 14911 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1632, + "step": 14912 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1538, + "step": 14913 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 14914 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.208, + "step": 14915 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 14916 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 14917 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1601, + "step": 14918 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 14919 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 14920 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1634, + "step": 14921 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 14922 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 14923 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 14924 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1754, + "step": 14925 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 14926 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 14927 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 14928 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.164, + "step": 14929 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1564, + "step": 14930 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 14931 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 14932 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 14933 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1787, + "step": 14934 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1878, + "step": 14935 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.2032, + "step": 14936 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 14937 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1674, + "step": 14938 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 14939 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1641, + "step": 14940 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1753, + "step": 14941 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 14942 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1628, + "step": 14943 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.2089, + "step": 14944 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 14945 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 14946 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 14947 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.2006, + "step": 14948 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 14949 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 14950 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.154, + "step": 14951 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 14952 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1735, + "step": 14953 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1579, + "step": 14954 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1459, + "step": 14955 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 14956 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1851, + "step": 14957 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1653, + "step": 14958 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1908, + "step": 14959 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 14960 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1592, + "step": 14961 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.176, + "step": 14962 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1497, + "step": 14963 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1573, + "step": 14964 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1534, + "step": 14965 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 14966 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1546, + "step": 14967 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 14968 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 14969 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1632, + "step": 14970 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 14971 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 14972 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1801, + "step": 14973 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 14974 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1642, + "step": 14975 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 14976 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 14977 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 14978 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1602, + "step": 14979 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1541, + "step": 14980 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 14981 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1794, + "step": 14982 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1804, + "step": 14983 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1943, + "step": 14984 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1887, + "step": 14985 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.2058, + "step": 14986 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 14987 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1666, + "step": 14988 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.2165, + "step": 14989 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 14990 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 14991 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.19, + "step": 14992 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 14993 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1729, + "step": 14994 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.148, + "step": 14995 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1589, + "step": 14996 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 14997 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1786, + "step": 14998 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.2925, + "step": 14999 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1774, + "step": 15000 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1527, + "step": 15001 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1686, + "step": 15002 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 15003 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1684, + "step": 15004 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1804, + "step": 15005 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 15006 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1478, + "step": 15007 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1588, + "step": 15008 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 15009 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1598, + "step": 15010 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1726, + "step": 15011 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.2163, + "step": 15012 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1804, + "step": 15013 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1748, + "step": 15014 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1896, + "step": 15015 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 15016 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 15017 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 15018 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 15019 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1812, + "step": 15020 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.154, + "step": 15021 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.172, + "step": 15022 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 15023 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1768, + "step": 15024 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1973, + "step": 15025 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 15026 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 15027 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1736, + "step": 15028 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 15029 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 15030 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 15031 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1805, + "step": 15032 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1627, + "step": 15033 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.188, + "step": 15034 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1972, + "step": 15035 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.162, + "step": 15036 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1649, + "step": 15037 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1851, + "step": 15038 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1803, + "step": 15039 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1704, + "step": 15040 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 15041 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1614, + "step": 15042 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1502, + "step": 15043 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1733, + "step": 15044 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.182, + "step": 15045 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1528, + "step": 15046 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 15047 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 15048 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.176, + "step": 15049 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.2012, + "step": 15050 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1803, + "step": 15051 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1885, + "step": 15052 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1697, + "step": 15053 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 15054 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 15055 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 15056 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1588, + "step": 15057 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.176, + "step": 15058 + }, + { + "epoch": 0.91, + "learning_rate": 2e-05, + "loss": 0.1659, + "step": 15059 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 15060 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 15061 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 15062 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 15063 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 15064 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 15065 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1811, + "step": 15066 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1524, + "step": 15067 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 15068 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 15069 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.2042, + "step": 15070 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1705, + "step": 15071 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 15072 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1691, + "step": 15073 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1702, + "step": 15074 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1674, + "step": 15075 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1715, + "step": 15076 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1809, + "step": 15077 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 15078 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 15079 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 15080 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 15081 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 15082 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 15083 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.2037, + "step": 15084 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 15085 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1739, + "step": 15086 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1555, + "step": 15087 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1701, + "step": 15088 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1709, + "step": 15089 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 15090 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 15091 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1576, + "step": 15092 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1676, + "step": 15093 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1503, + "step": 15094 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 15095 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1588, + "step": 15096 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1549, + "step": 15097 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 15098 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1429, + "step": 15099 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1933, + "step": 15100 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 15101 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 15102 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1585, + "step": 15103 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 15104 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 15105 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 15106 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 15107 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1745, + "step": 15108 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1707, + "step": 15109 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 15110 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1741, + "step": 15111 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 15112 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 15113 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 15114 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.2, + "step": 15115 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 15116 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1863, + "step": 15117 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1505, + "step": 15118 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1518, + "step": 15119 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1662, + "step": 15120 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1929, + "step": 15121 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1899, + "step": 15122 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1603, + "step": 15123 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 15124 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1801, + "step": 15125 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.2338, + "step": 15126 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 15127 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 15128 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 15129 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1646, + "step": 15130 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1655, + "step": 15131 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 15132 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1603, + "step": 15133 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 15134 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 15135 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 15136 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 15137 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 15138 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1763, + "step": 15139 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 15140 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.175, + "step": 15141 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1611, + "step": 15142 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1993, + "step": 15143 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1517, + "step": 15144 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1548, + "step": 15145 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 15146 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1742, + "step": 15147 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.2038, + "step": 15148 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 15149 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 15150 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 15151 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 15152 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.0834, + "step": 15153 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 15154 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 15155 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 15156 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.217, + "step": 15157 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1752, + "step": 15158 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.168, + "step": 15159 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1478, + "step": 15160 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1789, + "step": 15161 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 15162 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 15163 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 15164 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 15165 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1495, + "step": 15166 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1865, + "step": 15167 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 15168 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.181, + "step": 15169 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1608, + "step": 15170 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 15171 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1589, + "step": 15172 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 15173 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1755, + "step": 15174 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 15175 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1723, + "step": 15176 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1681, + "step": 15177 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1918, + "step": 15178 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 15179 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 15180 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 15181 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1526, + "step": 15182 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 15183 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 15184 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1853, + "step": 15185 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 15186 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1592, + "step": 15187 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 15188 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1628, + "step": 15189 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1933, + "step": 15190 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.16, + "step": 15191 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 15192 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1652, + "step": 15193 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 15194 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 15195 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 15196 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1743, + "step": 15197 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.2118, + "step": 15198 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 15199 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.2119, + "step": 15200 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1874, + "step": 15201 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1714, + "step": 15202 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 15203 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.2158, + "step": 15204 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 15205 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1791, + "step": 15206 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1871, + "step": 15207 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1665, + "step": 15208 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1952, + "step": 15209 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 15210 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 15211 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1517, + "step": 15212 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1687, + "step": 15213 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 15214 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1606, + "step": 15215 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1735, + "step": 15216 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1653, + "step": 15217 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.2066, + "step": 15218 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1723, + "step": 15219 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 15220 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 15221 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1734, + "step": 15222 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1562, + "step": 15223 + }, + { + "epoch": 0.92, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 15224 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 15225 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 15226 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1833, + "step": 15227 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1648, + "step": 15228 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1618, + "step": 15229 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1561, + "step": 15230 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 15231 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.196, + "step": 15232 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 15233 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1832, + "step": 15234 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 15235 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1618, + "step": 15236 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1889, + "step": 15237 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.16, + "step": 15238 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 15239 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 15240 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1781, + "step": 15241 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1756, + "step": 15242 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1778, + "step": 15243 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.2086, + "step": 15244 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 15245 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1585, + "step": 15246 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1655, + "step": 15247 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 15248 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1789, + "step": 15249 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1666, + "step": 15250 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 15251 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1462, + "step": 15252 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1712, + "step": 15253 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1573, + "step": 15254 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 15255 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1704, + "step": 15256 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1878, + "step": 15257 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1893, + "step": 15258 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 15259 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 15260 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1456, + "step": 15261 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1707, + "step": 15262 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1697, + "step": 15263 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1782, + "step": 15264 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1532, + "step": 15265 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 15266 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 15267 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1578, + "step": 15268 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 15269 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 15270 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 15271 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1768, + "step": 15272 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1781, + "step": 15273 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 15274 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1844, + "step": 15275 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1495, + "step": 15276 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1721, + "step": 15277 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1616, + "step": 15278 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1616, + "step": 15279 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 15280 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.2162, + "step": 15281 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1614, + "step": 15282 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1776, + "step": 15283 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 15284 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 15285 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 15286 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 15287 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 15288 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 15289 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 15290 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 15291 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 15292 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 15293 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.18, + "step": 15294 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 15295 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1627, + "step": 15296 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1485, + "step": 15297 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.176, + "step": 15298 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 15299 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1564, + "step": 15300 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1906, + "step": 15301 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.2006, + "step": 15302 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 15303 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 15304 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 15305 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.191, + "step": 15306 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1994, + "step": 15307 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.193, + "step": 15308 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 15309 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 15310 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.174, + "step": 15311 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1601, + "step": 15312 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1799, + "step": 15313 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 15314 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1574, + "step": 15315 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 15316 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 15317 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 15318 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 15319 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1541, + "step": 15320 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 15321 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 15322 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1974, + "step": 15323 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 15324 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 15325 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1976, + "step": 15326 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.224, + "step": 15327 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1566, + "step": 15328 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 15329 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.153, + "step": 15330 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 15331 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1754, + "step": 15332 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 15333 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 15334 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1998, + "step": 15335 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1568, + "step": 15336 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 15337 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1604, + "step": 15338 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 15339 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 15340 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.2237, + "step": 15341 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.2131, + "step": 15342 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1682, + "step": 15343 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1679, + "step": 15344 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 15345 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 15346 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 15347 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1617, + "step": 15348 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1701, + "step": 15349 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 15350 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1892, + "step": 15351 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.172, + "step": 15352 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1714, + "step": 15353 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 15354 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 15355 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 15356 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1793, + "step": 15357 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1734, + "step": 15358 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1899, + "step": 15359 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1588, + "step": 15360 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1992, + "step": 15361 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1625, + "step": 15362 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1677, + "step": 15363 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 15364 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1668, + "step": 15365 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1634, + "step": 15366 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1695, + "step": 15367 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 15368 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1715, + "step": 15369 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 15370 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.155, + "step": 15371 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 15372 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 15373 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 15374 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 15375 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1576, + "step": 15376 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1589, + "step": 15377 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1755, + "step": 15378 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 15379 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1699, + "step": 15380 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.2207, + "step": 15381 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 15382 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 15383 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 15384 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 15385 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 15386 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.2208, + "step": 15387 + }, + { + "epoch": 0.93, + "learning_rate": 2e-05, + "loss": 0.1589, + "step": 15388 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.2029, + "step": 15389 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 15390 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1602, + "step": 15391 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1684, + "step": 15392 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 15393 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 15394 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1621, + "step": 15395 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 15396 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1657, + "step": 15397 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 15398 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 15399 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 15400 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 15401 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 15402 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 15403 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1552, + "step": 15404 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1813, + "step": 15405 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1664, + "step": 15406 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 15407 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1938, + "step": 15408 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 15409 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 15410 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 15411 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1774, + "step": 15412 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 15413 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.2115, + "step": 15414 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 15415 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1499, + "step": 15416 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 15417 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1526, + "step": 15418 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.155, + "step": 15419 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 15420 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1925, + "step": 15421 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1945, + "step": 15422 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1552, + "step": 15423 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 15424 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1764, + "step": 15425 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 15426 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.2017, + "step": 15427 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 15428 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.2039, + "step": 15429 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 15430 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1665, + "step": 15431 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1698, + "step": 15432 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 15433 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 15434 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 15435 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1804, + "step": 15436 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.163, + "step": 15437 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1462, + "step": 15438 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 15439 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 15440 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 15441 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1717, + "step": 15442 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1524, + "step": 15443 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1906, + "step": 15444 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1674, + "step": 15445 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1896, + "step": 15446 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 15447 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 15448 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1947, + "step": 15449 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.162, + "step": 15450 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1459, + "step": 15451 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1845, + "step": 15452 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 15453 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1567, + "step": 15454 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.148, + "step": 15455 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1719, + "step": 15456 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 15457 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.182, + "step": 15458 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 15459 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1632, + "step": 15460 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.198, + "step": 15461 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1553, + "step": 15462 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 15463 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1862, + "step": 15464 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1667, + "step": 15465 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1648, + "step": 15466 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 15467 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1552, + "step": 15468 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 15469 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 15470 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 15471 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1648, + "step": 15472 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 15473 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.2126, + "step": 15474 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1744, + "step": 15475 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1672, + "step": 15476 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1722, + "step": 15477 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1756, + "step": 15478 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1562, + "step": 15479 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.168, + "step": 15480 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1502, + "step": 15481 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1807, + "step": 15482 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1552, + "step": 15483 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 15484 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1607, + "step": 15485 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1704, + "step": 15486 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 15487 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1554, + "step": 15488 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1965, + "step": 15489 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1655, + "step": 15490 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 15491 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 15492 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1462, + "step": 15493 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1762, + "step": 15494 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1723, + "step": 15495 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1638, + "step": 15496 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1812, + "step": 15497 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1562, + "step": 15498 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 15499 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 15500 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1861, + "step": 15501 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1693, + "step": 15502 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 15503 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 15504 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 15505 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 15506 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1705, + "step": 15507 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.173, + "step": 15508 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 15509 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 15510 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1602, + "step": 15511 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 15512 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1883, + "step": 15513 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1694, + "step": 15514 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1739, + "step": 15515 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1971, + "step": 15516 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 15517 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1596, + "step": 15518 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1979, + "step": 15519 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.2094, + "step": 15520 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 15521 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1563, + "step": 15522 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1823, + "step": 15523 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 15524 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 15525 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1722, + "step": 15526 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 15527 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.2114, + "step": 15528 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.162, + "step": 15529 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.2249, + "step": 15530 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.2076, + "step": 15531 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1628, + "step": 15532 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1762, + "step": 15533 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 15534 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1601, + "step": 15535 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 15536 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 15537 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.175, + "step": 15538 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1686, + "step": 15539 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 15540 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.189, + "step": 15541 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 15542 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 15543 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1629, + "step": 15544 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1827, + "step": 15545 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 15546 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 15547 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 15548 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 15549 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.2085, + "step": 15550 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 15551 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1909, + "step": 15552 + }, + { + "epoch": 0.94, + "learning_rate": 2e-05, + "loss": 0.1534, + "step": 15553 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1746, + "step": 15554 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1505, + "step": 15555 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1887, + "step": 15556 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1881, + "step": 15557 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.173, + "step": 15558 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 15559 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 15560 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 15561 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 15562 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 15563 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1621, + "step": 15564 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 15565 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1882, + "step": 15566 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 15567 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1726, + "step": 15568 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1663, + "step": 15569 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1778, + "step": 15570 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1747, + "step": 15571 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 15572 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.196, + "step": 15573 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1921, + "step": 15574 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1735, + "step": 15575 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1553, + "step": 15576 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 15577 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1552, + "step": 15578 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 15579 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1694, + "step": 15580 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 15581 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 15582 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 15583 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 15584 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 15585 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1717, + "step": 15586 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 15587 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1909, + "step": 15588 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 15589 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1819, + "step": 15590 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1724, + "step": 15591 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 15592 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 15593 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1647, + "step": 15594 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 15595 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 15596 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 15597 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 15598 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 15599 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 15600 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1453, + "step": 15601 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.2162, + "step": 15602 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 15603 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1626, + "step": 15604 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 15605 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 15606 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 15607 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1498, + "step": 15608 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1769, + "step": 15609 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 15610 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.161, + "step": 15611 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 15612 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1612, + "step": 15613 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 15614 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1706, + "step": 15615 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 15616 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 15617 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1705, + "step": 15618 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 15619 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1541, + "step": 15620 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1566, + "step": 15621 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 15622 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1554, + "step": 15623 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 15624 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1918, + "step": 15625 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1612, + "step": 15626 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 15627 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1633, + "step": 15628 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1567, + "step": 15629 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1641, + "step": 15630 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1711, + "step": 15631 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 15632 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1921, + "step": 15633 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 15634 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 15635 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1728, + "step": 15636 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 15637 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 15638 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1562, + "step": 15639 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1698, + "step": 15640 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1639, + "step": 15641 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 15642 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1558, + "step": 15643 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 15644 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1588, + "step": 15645 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1984, + "step": 15646 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1469, + "step": 15647 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 15648 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 15649 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1874, + "step": 15650 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 15651 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1734, + "step": 15652 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 15653 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 15654 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.2162, + "step": 15655 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 15656 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 15657 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 15658 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 15659 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 15660 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1459, + "step": 15661 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1744, + "step": 15662 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1747, + "step": 15663 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 15664 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1831, + "step": 15665 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1688, + "step": 15666 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1961, + "step": 15667 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1783, + "step": 15668 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1996, + "step": 15669 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 15670 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 15671 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 15672 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1936, + "step": 15673 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1867, + "step": 15674 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 15675 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 15676 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1498, + "step": 15677 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1817, + "step": 15678 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1648, + "step": 15679 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 15680 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1538, + "step": 15681 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1861, + "step": 15682 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1611, + "step": 15683 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.206, + "step": 15684 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1865, + "step": 15685 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1697, + "step": 15686 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1629, + "step": 15687 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 15688 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1599, + "step": 15689 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 15690 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 15691 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1707, + "step": 15692 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1607, + "step": 15693 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.18, + "step": 15694 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1918, + "step": 15695 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 15696 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 15697 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1971, + "step": 15698 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 15699 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 15700 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 15701 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.2007, + "step": 15702 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 15703 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1638, + "step": 15704 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 15705 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 15706 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1662, + "step": 15707 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1908, + "step": 15708 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 15709 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 15710 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.2027, + "step": 15711 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 15712 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1986, + "step": 15713 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.215, + "step": 15714 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1603, + "step": 15715 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 15716 + }, + { + "epoch": 0.95, + "learning_rate": 2e-05, + "loss": 0.1647, + "step": 15717 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 15718 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.17, + "step": 15719 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.196, + "step": 15720 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 15721 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1844, + "step": 15722 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1734, + "step": 15723 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1746, + "step": 15724 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.2059, + "step": 15725 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1842, + "step": 15726 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 15727 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1677, + "step": 15728 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.198, + "step": 15729 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 15730 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 15731 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 15732 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 15733 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 15734 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 15735 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 15736 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1823, + "step": 15737 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 15738 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 15739 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 15740 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1517, + "step": 15741 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 15742 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 15743 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 15744 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 15745 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 15746 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 15747 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.152, + "step": 15748 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1969, + "step": 15749 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1651, + "step": 15750 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 15751 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1922, + "step": 15752 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 15753 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 15754 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 15755 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1734, + "step": 15756 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 15757 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1676, + "step": 15758 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1754, + "step": 15759 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 15760 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 15761 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1677, + "step": 15762 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1961, + "step": 15763 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 15764 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 15765 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 15766 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 15767 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 15768 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 15769 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 15770 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.2069, + "step": 15771 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 15772 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1582, + "step": 15773 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 15774 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 15775 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 15776 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 15777 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 15778 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1696, + "step": 15779 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1478, + "step": 15780 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1685, + "step": 15781 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 15782 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 15783 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 15784 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1775, + "step": 15785 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 15786 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 15787 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 15788 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.2094, + "step": 15789 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.2033, + "step": 15790 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1869, + "step": 15791 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 15792 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 15793 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 15794 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1892, + "step": 15795 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 15796 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1909, + "step": 15797 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1642, + "step": 15798 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 15799 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1804, + "step": 15800 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 15801 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1737, + "step": 15802 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 15803 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1818, + "step": 15804 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 15805 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.159, + "step": 15806 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.155, + "step": 15807 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.169, + "step": 15808 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1695, + "step": 15809 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1608, + "step": 15810 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1635, + "step": 15811 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 15812 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.183, + "step": 15813 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1628, + "step": 15814 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 15815 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1618, + "step": 15816 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 15817 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1533, + "step": 15818 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1659, + "step": 15819 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1723, + "step": 15820 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 15821 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 15822 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1462, + "step": 15823 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1677, + "step": 15824 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 15825 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1775, + "step": 15826 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 15827 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1745, + "step": 15828 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 15829 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1602, + "step": 15830 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1557, + "step": 15831 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1585, + "step": 15832 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1585, + "step": 15833 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1519, + "step": 15834 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.147, + "step": 15835 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.2108, + "step": 15836 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1719, + "step": 15837 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 15838 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 15839 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1546, + "step": 15840 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 15841 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.2095, + "step": 15842 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1597, + "step": 15843 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1618, + "step": 15844 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.156, + "step": 15845 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1716, + "step": 15846 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1599, + "step": 15847 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 15848 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 15849 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1758, + "step": 15850 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1787, + "step": 15851 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 15852 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1974, + "step": 15853 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 15854 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1564, + "step": 15855 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1497, + "step": 15856 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.2012, + "step": 15857 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.2025, + "step": 15858 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 15859 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 15860 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1612, + "step": 15861 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 15862 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1614, + "step": 15863 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 15864 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.2068, + "step": 15865 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1924, + "step": 15866 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1737, + "step": 15867 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 15868 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1871, + "step": 15869 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1515, + "step": 15870 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 15871 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 15872 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1629, + "step": 15873 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.178, + "step": 15874 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.185, + "step": 15875 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.168, + "step": 15876 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 15877 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 15878 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1527, + "step": 15879 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 15880 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 15881 + }, + { + "epoch": 0.96, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 15882 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1554, + "step": 15883 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1839, + "step": 15884 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1654, + "step": 15885 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 15886 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.2006, + "step": 15887 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 15888 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 15889 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 15890 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 15891 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 15892 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1888, + "step": 15893 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1459, + "step": 15894 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1789, + "step": 15895 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 15896 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 15897 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1676, + "step": 15898 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1451, + "step": 15899 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 15900 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1605, + "step": 15901 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.146, + "step": 15902 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1876, + "step": 15903 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 15904 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1551, + "step": 15905 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1923, + "step": 15906 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1727, + "step": 15907 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 15908 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1616, + "step": 15909 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1966, + "step": 15910 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1459, + "step": 15911 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 15912 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1787, + "step": 15913 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1551, + "step": 15914 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 15915 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 15916 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 15917 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 15918 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.2163, + "step": 15919 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 15920 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1796, + "step": 15921 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1746, + "step": 15922 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1549, + "step": 15923 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1611, + "step": 15924 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 15925 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 15926 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 15927 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 15928 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 15929 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1596, + "step": 15930 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.17, + "step": 15931 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.2081, + "step": 15932 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 15933 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 15934 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 15935 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.165, + "step": 15936 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 15937 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1721, + "step": 15938 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 15939 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1462, + "step": 15940 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 15941 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 15942 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1804, + "step": 15943 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1447, + "step": 15944 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1718, + "step": 15945 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1689, + "step": 15946 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 15947 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 15948 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1777, + "step": 15949 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.2039, + "step": 15950 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 15951 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 15952 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 15953 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 15954 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1531, + "step": 15955 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 15956 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 15957 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1737, + "step": 15958 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 15959 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 15960 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1557, + "step": 15961 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1874, + "step": 15962 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 15963 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 15964 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 15965 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 15966 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 15967 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.2026, + "step": 15968 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 15969 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.177, + "step": 15970 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 15971 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 15972 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1754, + "step": 15973 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 15974 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 15975 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1462, + "step": 15976 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 15977 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1702, + "step": 15978 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.2249, + "step": 15979 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 15980 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 15981 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 15982 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1617, + "step": 15983 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 15984 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 15985 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 15986 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 15987 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 15988 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1984, + "step": 15989 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1772, + "step": 15990 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 15991 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1847, + "step": 15992 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1663, + "step": 15993 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1806, + "step": 15994 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1633, + "step": 15995 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 15996 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1735, + "step": 15997 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1694, + "step": 15998 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 15999 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1589, + "step": 16000 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 16001 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1629, + "step": 16002 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 16003 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.2192, + "step": 16004 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 16005 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 16006 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 16007 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1668, + "step": 16008 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.2081, + "step": 16009 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1699, + "step": 16010 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1895, + "step": 16011 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1468, + "step": 16012 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 16013 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1928, + "step": 16014 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1613, + "step": 16015 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 16016 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 16017 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 16018 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 16019 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.2183, + "step": 16020 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1651, + "step": 16021 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1849, + "step": 16022 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1495, + "step": 16023 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1717, + "step": 16024 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 16025 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1538, + "step": 16026 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1925, + "step": 16027 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 16028 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 16029 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 16030 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 16031 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.2246, + "step": 16032 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1535, + "step": 16033 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1641, + "step": 16034 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1552, + "step": 16035 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1741, + "step": 16036 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1516, + "step": 16037 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1505, + "step": 16038 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1538, + "step": 16039 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 16040 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1597, + "step": 16041 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1541, + "step": 16042 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 16043 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1498, + "step": 16044 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1459, + "step": 16045 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 16046 + }, + { + "epoch": 0.97, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 16047 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 16048 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1601, + "step": 16049 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1844, + "step": 16050 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1647, + "step": 16051 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1669, + "step": 16052 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 16053 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1815, + "step": 16054 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.2058, + "step": 16055 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1622, + "step": 16056 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 16057 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1746, + "step": 16058 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 16059 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1586, + "step": 16060 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 16061 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1488, + "step": 16062 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 16063 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 16064 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1893, + "step": 16065 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1953, + "step": 16066 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 16067 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 16068 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.185, + "step": 16069 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1595, + "step": 16070 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 16071 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 16072 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1541, + "step": 16073 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 16074 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 16075 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 16076 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1684, + "step": 16077 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 16078 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 16079 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 16080 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 16081 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 16082 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1532, + "step": 16083 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 16084 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 16085 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1617, + "step": 16086 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1664, + "step": 16087 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 16088 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 16089 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 16090 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 16091 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1794, + "step": 16092 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 16093 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1814, + "step": 16094 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1937, + "step": 16095 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1724, + "step": 16096 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 16097 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1648, + "step": 16098 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1786, + "step": 16099 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.2126, + "step": 16100 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 16101 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 16102 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.205, + "step": 16103 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 16104 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1885, + "step": 16105 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1654, + "step": 16106 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1555, + "step": 16107 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1469, + "step": 16108 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1586, + "step": 16109 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 16110 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1729, + "step": 16111 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 16112 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1495, + "step": 16113 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1802, + "step": 16114 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 16115 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 16116 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 16117 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1808, + "step": 16118 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 16119 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 16120 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1625, + "step": 16121 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 16122 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1632, + "step": 16123 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 16124 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 16125 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1613, + "step": 16126 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 16127 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1657, + "step": 16128 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1817, + "step": 16129 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1731, + "step": 16130 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 16131 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 16132 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1715, + "step": 16133 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 16134 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1974, + "step": 16135 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1526, + "step": 16136 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 16137 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1629, + "step": 16138 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1712, + "step": 16139 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1635, + "step": 16140 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 16141 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 16142 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 16143 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1564, + "step": 16144 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.2098, + "step": 16145 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1811, + "step": 16146 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1673, + "step": 16147 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.2156, + "step": 16148 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 16149 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 16150 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.2058, + "step": 16151 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.147, + "step": 16152 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 16153 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1591, + "step": 16154 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 16155 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1884, + "step": 16156 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 16157 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1676, + "step": 16158 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1748, + "step": 16159 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1626, + "step": 16160 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1686, + "step": 16161 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.2339, + "step": 16162 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 16163 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1989, + "step": 16164 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 16165 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 16166 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 16167 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 16168 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 16169 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 16170 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1953, + "step": 16171 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1751, + "step": 16172 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1645, + "step": 16173 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 16174 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1821, + "step": 16175 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1443, + "step": 16176 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 16177 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 16178 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1864, + "step": 16179 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 16180 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 16181 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 16182 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 16183 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 16184 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1856, + "step": 16185 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 16186 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 16187 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 16188 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 16189 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 16190 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 16191 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 16192 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.178, + "step": 16193 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 16194 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 16195 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1562, + "step": 16196 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 16197 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1607, + "step": 16198 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1898, + "step": 16199 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.2096, + "step": 16200 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1657, + "step": 16201 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1647, + "step": 16202 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1692, + "step": 16203 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1735, + "step": 16204 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1627, + "step": 16205 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 16206 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 16207 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 16208 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.2465, + "step": 16209 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1815, + "step": 16210 + }, + { + "epoch": 0.98, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 16211 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 16212 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1991, + "step": 16213 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1566, + "step": 16214 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1715, + "step": 16215 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1612, + "step": 16216 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1498, + "step": 16217 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1598, + "step": 16218 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1827, + "step": 16219 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 16220 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.249, + "step": 16221 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1804, + "step": 16222 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1899, + "step": 16223 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 16224 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1527, + "step": 16225 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 16226 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 16227 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 16228 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1857, + "step": 16229 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 16230 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.216, + "step": 16231 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 16232 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 16233 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1706, + "step": 16234 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1643, + "step": 16235 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 16236 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1668, + "step": 16237 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1553, + "step": 16238 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1733, + "step": 16239 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1459, + "step": 16240 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 16241 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.222, + "step": 16242 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1807, + "step": 16243 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 16244 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 16245 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1557, + "step": 16246 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 16247 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1708, + "step": 16248 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.147, + "step": 16249 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 16250 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 16251 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 16252 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 16253 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 16254 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1695, + "step": 16255 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 16256 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 16257 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1885, + "step": 16258 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1648, + "step": 16259 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 16260 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 16261 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 16262 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 16263 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 16264 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1611, + "step": 16265 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1913, + "step": 16266 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1451, + "step": 16267 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 16268 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1978, + "step": 16269 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 16270 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1525, + "step": 16271 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1469, + "step": 16272 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 16273 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1669, + "step": 16274 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1853, + "step": 16275 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1934, + "step": 16276 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1735, + "step": 16277 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 16278 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1542, + "step": 16279 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 16280 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 16281 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 16282 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1927, + "step": 16283 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1756, + "step": 16284 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1499, + "step": 16285 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1603, + "step": 16286 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 16287 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 16288 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.2211, + "step": 16289 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1976, + "step": 16290 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1908, + "step": 16291 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 16292 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1732, + "step": 16293 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 16294 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 16295 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 16296 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1809, + "step": 16297 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1713, + "step": 16298 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 16299 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1505, + "step": 16300 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1499, + "step": 16301 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 16302 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.148, + "step": 16303 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1624, + "step": 16304 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1468, + "step": 16305 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 16306 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 16307 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 16308 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1961, + "step": 16309 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.2078, + "step": 16310 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1884, + "step": 16311 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1599, + "step": 16312 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 16313 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 16314 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 16315 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1821, + "step": 16316 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1764, + "step": 16317 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 16318 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1699, + "step": 16319 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 16320 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.2242, + "step": 16321 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1804, + "step": 16322 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 16323 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1502, + "step": 16324 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 16325 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.248, + "step": 16326 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 16327 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 16328 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1602, + "step": 16329 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 16330 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 16331 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 16332 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 16333 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1521, + "step": 16334 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 16335 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1911, + "step": 16336 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 16337 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 16338 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 16339 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 16340 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.153, + "step": 16341 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1606, + "step": 16342 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1591, + "step": 16343 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 16344 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 16345 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1476, + "step": 16346 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1465, + "step": 16347 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.155, + "step": 16348 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1702, + "step": 16349 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1929, + "step": 16350 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 16351 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 16352 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1582, + "step": 16353 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 16354 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1625, + "step": 16355 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.155, + "step": 16356 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1946, + "step": 16357 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.2057, + "step": 16358 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1633, + "step": 16359 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1557, + "step": 16360 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1966, + "step": 16361 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 16362 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.16, + "step": 16363 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 16364 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1588, + "step": 16365 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 16366 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1648, + "step": 16367 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 16368 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1637, + "step": 16369 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 16370 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 16371 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1696, + "step": 16372 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1766, + "step": 16373 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1777, + "step": 16374 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1542, + "step": 16375 + }, + { + "epoch": 0.99, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 16376 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1779, + "step": 16377 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1453, + "step": 16378 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1683, + "step": 16379 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 16380 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 16381 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1558, + "step": 16382 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 16383 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1996, + "step": 16384 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 16385 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 16386 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.206, + "step": 16387 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 16388 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1653, + "step": 16389 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.156, + "step": 16390 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1453, + "step": 16391 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 16392 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1476, + "step": 16393 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1468, + "step": 16394 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.2037, + "step": 16395 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 16396 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 16397 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1517, + "step": 16398 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1705, + "step": 16399 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 16400 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 16401 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1599, + "step": 16402 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1735, + "step": 16403 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 16404 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.144, + "step": 16405 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 16406 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.183, + "step": 16407 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 16408 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 16409 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 16410 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 16411 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 16412 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1726, + "step": 16413 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 16414 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1888, + "step": 16415 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.158, + "step": 16416 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1725, + "step": 16417 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 16418 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 16419 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1684, + "step": 16420 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 16421 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 16422 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 16423 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1453, + "step": 16424 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.175, + "step": 16425 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 16426 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 16427 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1702, + "step": 16428 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.2125, + "step": 16429 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 16430 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 16431 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 16432 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 16433 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1942, + "step": 16434 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1656, + "step": 16435 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.2166, + "step": 16436 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1538, + "step": 16437 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1705, + "step": 16438 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 16439 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 16440 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 16441 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1715, + "step": 16442 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1849, + "step": 16443 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1488, + "step": 16444 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 16445 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 16446 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1651, + "step": 16447 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.2037, + "step": 16448 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1938, + "step": 16449 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 16450 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1738, + "step": 16451 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 16452 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 16453 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1503, + "step": 16454 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1772, + "step": 16455 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 16456 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1675, + "step": 16457 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 16458 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 16459 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 16460 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1591, + "step": 16461 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1684, + "step": 16462 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 16463 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 16464 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 16465 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 16466 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 16467 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 16468 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 16469 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.205, + "step": 16470 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1719, + "step": 16471 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 16472 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1639, + "step": 16473 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 16474 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 16475 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 16476 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 16477 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1593, + "step": 16478 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 16479 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 16480 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 16481 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1541, + "step": 16482 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 16483 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1644, + "step": 16484 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.163, + "step": 16485 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 16486 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1633, + "step": 16487 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 16488 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 16489 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1525, + "step": 16490 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 16491 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 16492 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 16493 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 16494 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 16495 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1702, + "step": 16496 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 16497 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 16498 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 16499 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 16500 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 16501 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 16502 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 16503 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1478, + "step": 16504 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1676, + "step": 16505 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 16506 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1628, + "step": 16507 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 16508 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 16509 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 16510 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.145, + "step": 16511 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 16512 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 16513 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 16514 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 16515 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1764, + "step": 16516 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 16517 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 16518 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1505, + "step": 16519 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1731, + "step": 16520 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 16521 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 16522 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 16523 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 16524 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 16525 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 16526 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 16527 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 16528 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 16529 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 16530 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 16531 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 16532 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1839, + "step": 16533 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1495, + "step": 16534 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1675, + "step": 16535 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 16536 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 16537 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 16538 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 16539 + }, + { + "epoch": 1.0, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 16540 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 16541 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1527, + "step": 16542 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 16543 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 16544 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 16545 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 16546 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1511, + "step": 16547 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 16548 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1646, + "step": 16549 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 16550 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1564, + "step": 16551 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1526, + "step": 16552 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 16553 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 16554 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 16555 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 16556 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 16557 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 16558 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 16559 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 16560 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 16561 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 16562 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 16563 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 16564 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1945, + "step": 16565 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 16566 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 16567 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 16568 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1641, + "step": 16569 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 16570 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 16571 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 16572 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 16573 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 16574 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 16575 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 16576 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 16577 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 16578 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 16579 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1591, + "step": 16580 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 16581 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 16582 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 16583 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 16584 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 16585 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 16586 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 16587 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1527, + "step": 16588 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.161, + "step": 16589 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 16590 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 16591 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 16592 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 16593 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 16594 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 16595 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 16596 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1586, + "step": 16597 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 16598 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 16599 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 16600 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 16601 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 16602 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 16603 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 16604 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 16605 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 16606 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 16607 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 16608 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1598, + "step": 16609 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.146, + "step": 16610 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 16611 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1643, + "step": 16612 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 16613 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 16614 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1982, + "step": 16615 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 16616 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 16617 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 16618 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1715, + "step": 16619 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 16620 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 16621 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 16622 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 16623 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 16624 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.169, + "step": 16625 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 16626 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 16627 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 16628 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 16629 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.194, + "step": 16630 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 16631 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 16632 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1429, + "step": 16633 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 16634 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 16635 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 16636 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 16637 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.2108, + "step": 16638 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1521, + "step": 16639 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 16640 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1659, + "step": 16641 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 16642 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 16643 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 16644 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 16645 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1538, + "step": 16646 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1768, + "step": 16647 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 16648 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 16649 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1534, + "step": 16650 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1798, + "step": 16651 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 16652 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 16653 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 16654 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 16655 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 16656 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 16657 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 16658 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 16659 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 16660 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1647, + "step": 16661 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1546, + "step": 16662 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 16663 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 16664 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1622, + "step": 16665 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1829, + "step": 16666 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1583, + "step": 16667 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 16668 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 16669 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.171, + "step": 16670 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 16671 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 16672 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 16673 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1635, + "step": 16674 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 16675 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 16676 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 16677 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 16678 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 16679 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1618, + "step": 16680 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 16681 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 16682 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 16683 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 16684 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1468, + "step": 16685 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 16686 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 16687 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 16688 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 16689 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 16690 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 16691 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 16692 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 16693 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.0916, + "step": 16694 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 16695 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 16696 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 16697 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 16698 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 16699 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1654, + "step": 16700 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 16701 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 16702 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 16703 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 16704 + }, + { + "epoch": 1.01, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 16705 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 16706 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 16707 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.2051, + "step": 16708 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.153, + "step": 16709 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 16710 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1704, + "step": 16711 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 16712 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1817, + "step": 16713 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 16714 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 16715 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1623, + "step": 16716 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1497, + "step": 16717 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 16718 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 16719 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1563, + "step": 16720 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1782, + "step": 16721 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1606, + "step": 16722 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 16723 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 16724 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 16725 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 16726 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1749, + "step": 16727 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 16728 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1712, + "step": 16729 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1528, + "step": 16730 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 16731 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 16732 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 16733 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 16734 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 16735 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1561, + "step": 16736 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.175, + "step": 16737 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1671, + "step": 16738 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1582, + "step": 16739 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1599, + "step": 16740 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 16741 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 16742 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 16743 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 16744 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1635, + "step": 16745 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 16746 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 16747 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 16748 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 16749 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 16750 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1695, + "step": 16751 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 16752 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 16753 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 16754 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 16755 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 16756 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 16757 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 16758 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 16759 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 16760 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 16761 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 16762 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 16763 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 16764 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 16765 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 16766 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 16767 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 16768 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 16769 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 16770 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 16771 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 16772 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1524, + "step": 16773 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 16774 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 16775 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1849, + "step": 16776 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 16777 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 16778 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 16779 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1576, + "step": 16780 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 16781 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1658, + "step": 16782 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 16783 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 16784 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1576, + "step": 16785 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 16786 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 16787 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 16788 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.0917, + "step": 16789 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 16790 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 16791 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 16792 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 16793 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1538, + "step": 16794 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 16795 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 16796 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 16797 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 16798 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 16799 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 16800 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 16801 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 16802 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 16803 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 16804 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 16805 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1602, + "step": 16806 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 16807 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 16808 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 16809 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 16810 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 16811 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 16812 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 16813 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 16814 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 16815 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 16816 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 16817 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 16818 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 16819 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 16820 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 16821 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 16822 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1566, + "step": 16823 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 16824 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 16825 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 16826 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 16827 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 16828 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 16829 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.152, + "step": 16830 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1577, + "step": 16831 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 16832 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 16833 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 16834 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 16835 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1907, + "step": 16836 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 16837 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 16838 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1476, + "step": 16839 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 16840 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 16841 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 16842 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 16843 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 16844 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 16845 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1743, + "step": 16846 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1659, + "step": 16847 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 16848 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 16849 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1563, + "step": 16850 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 16851 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1596, + "step": 16852 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 16853 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.145, + "step": 16854 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1646, + "step": 16855 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1647, + "step": 16856 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 16857 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 16858 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 16859 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1896, + "step": 16860 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 16861 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 16862 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 16863 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1604, + "step": 16864 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 16865 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 16866 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 16867 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 16868 + }, + { + "epoch": 1.02, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 16869 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 16870 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 16871 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 16872 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 16873 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 16874 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 16875 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 16876 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 16877 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 16878 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1628, + "step": 16879 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 16880 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 16881 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 16882 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 16883 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 16884 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1687, + "step": 16885 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1531, + "step": 16886 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 16887 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 16888 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 16889 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.159, + "step": 16890 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 16891 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 16892 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.176, + "step": 16893 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 16894 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 16895 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1785, + "step": 16896 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1533, + "step": 16897 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.146, + "step": 16898 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1869, + "step": 16899 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 16900 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.164, + "step": 16901 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 16902 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 16903 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1698, + "step": 16904 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.2051, + "step": 16905 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1614, + "step": 16906 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 16907 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 16908 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 16909 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 16910 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1626, + "step": 16911 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 16912 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 16913 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 16914 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1793, + "step": 16915 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 16916 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.145, + "step": 16917 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 16918 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 16919 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 16920 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 16921 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 16922 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 16923 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 16924 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 16925 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 16926 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 16927 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 16928 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 16929 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 16930 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 16931 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1687, + "step": 16932 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 16933 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 16934 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 16935 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1624, + "step": 16936 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1802, + "step": 16937 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 16938 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 16939 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 16940 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 16941 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 16942 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1567, + "step": 16943 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 16944 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 16945 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 16946 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1986, + "step": 16947 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1739, + "step": 16948 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 16949 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 16950 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 16951 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 16952 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 16953 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 16954 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 16955 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1476, + "step": 16956 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 16957 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 16958 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1968, + "step": 16959 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1459, + "step": 16960 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1756, + "step": 16961 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1627, + "step": 16962 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 16963 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 16964 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 16965 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1817, + "step": 16966 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 16967 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 16968 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1563, + "step": 16969 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1911, + "step": 16970 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 16971 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1687, + "step": 16972 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1691, + "step": 16973 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1717, + "step": 16974 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 16975 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 16976 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 16977 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 16978 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1659, + "step": 16979 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 16980 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 16981 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 16982 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 16983 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1631, + "step": 16984 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 16985 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 16986 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 16987 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 16988 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 16989 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 16990 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 16991 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 16992 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1963, + "step": 16993 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 16994 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 16995 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 16996 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1592, + "step": 16997 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 16998 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 16999 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 17000 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1825, + "step": 17001 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1525, + "step": 17002 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 17003 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1598, + "step": 17004 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1794, + "step": 17005 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 17006 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 17007 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 17008 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 17009 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 17010 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.146, + "step": 17011 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 17012 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 17013 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1629, + "step": 17014 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 17015 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 17016 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 17017 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 17018 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 17019 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1632, + "step": 17020 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1541, + "step": 17021 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 17022 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 17023 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 17024 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1644, + "step": 17025 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1468, + "step": 17026 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 17027 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 17028 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 17029 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 17030 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 17031 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 17032 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 17033 + }, + { + "epoch": 1.03, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 17034 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 17035 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 17036 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1786, + "step": 17037 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 17038 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1591, + "step": 17039 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 17040 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 17041 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 17042 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 17043 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1498, + "step": 17044 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 17045 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 17046 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 17047 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 17048 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 17049 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1657, + "step": 17050 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 17051 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 17052 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 17053 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1844, + "step": 17054 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1678, + "step": 17055 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 17056 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 17057 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 17058 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 17059 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 17060 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 17061 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 17062 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1527, + "step": 17063 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 17064 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1499, + "step": 17065 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 17066 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 17067 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 17068 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 17069 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 17070 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 17071 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 17072 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 17073 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.144, + "step": 17074 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 17075 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1655, + "step": 17076 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 17077 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 17078 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 17079 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1766, + "step": 17080 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 17081 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 17082 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 17083 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 17084 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 17085 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 17086 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 17087 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1659, + "step": 17088 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 17089 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 17090 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 17091 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 17092 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 17093 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 17094 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1521, + "step": 17095 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 17096 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 17097 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 17098 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 17099 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 17100 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 17101 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 17102 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 17103 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1648, + "step": 17104 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 17105 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1515, + "step": 17106 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 17107 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 17108 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 17109 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 17110 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 17111 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.0875, + "step": 17112 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 17113 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 17114 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1803, + "step": 17115 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 17116 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1773, + "step": 17117 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 17118 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 17119 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1642, + "step": 17120 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 17121 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 17122 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1775, + "step": 17123 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 17124 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 17125 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 17126 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 17127 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 17128 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.16, + "step": 17129 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1589, + "step": 17130 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 17131 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.168, + "step": 17132 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 17133 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1791, + "step": 17134 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 17135 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 17136 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 17137 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1527, + "step": 17138 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 17139 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1451, + "step": 17140 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 17141 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 17142 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 17143 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1517, + "step": 17144 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1638, + "step": 17145 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 17146 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 17147 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1981, + "step": 17148 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 17149 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 17150 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1555, + "step": 17151 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 17152 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 17153 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 17154 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 17155 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1538, + "step": 17156 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 17157 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 17158 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1731, + "step": 17159 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1601, + "step": 17160 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 17161 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 17162 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 17163 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1644, + "step": 17164 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 17165 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 17166 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 17167 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 17168 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 17169 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1542, + "step": 17170 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 17171 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 17172 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 17173 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1714, + "step": 17174 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 17175 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 17176 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 17177 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 17178 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 17179 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 17180 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 17181 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 17182 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 17183 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1707, + "step": 17184 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 17185 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 17186 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 17187 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1453, + "step": 17188 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 17189 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 17190 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 17191 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 17192 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 17193 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 17194 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 17195 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1657, + "step": 17196 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 17197 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 17198 + }, + { + "epoch": 1.04, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 17199 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1847, + "step": 17200 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 17201 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 17202 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 17203 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 17204 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 17205 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 17206 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 17207 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 17208 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1737, + "step": 17209 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 17210 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 17211 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 17212 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 17213 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 17214 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.144, + "step": 17215 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 17216 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1639, + "step": 17217 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 17218 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 17219 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 17220 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 17221 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 17222 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 17223 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 17224 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 17225 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 17226 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 17227 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 17228 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 17229 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 17230 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 17231 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1821, + "step": 17232 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 17233 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 17234 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 17235 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 17236 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1848, + "step": 17237 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1447, + "step": 17238 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.16, + "step": 17239 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.18, + "step": 17240 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 17241 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 17242 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1696, + "step": 17243 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.2138, + "step": 17244 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1804, + "step": 17245 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 17246 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 17247 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 17248 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 17249 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 17250 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 17251 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 17252 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 17253 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 17254 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 17255 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 17256 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 17257 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1964, + "step": 17258 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 17259 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1597, + "step": 17260 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 17261 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.153, + "step": 17262 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1858, + "step": 17263 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1631, + "step": 17264 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1592, + "step": 17265 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1645, + "step": 17266 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 17267 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 17268 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1769, + "step": 17269 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 17270 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 17271 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 17272 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 17273 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 17274 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 17275 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1617, + "step": 17276 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 17277 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1687, + "step": 17278 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 17279 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 17280 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1585, + "step": 17281 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1644, + "step": 17282 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 17283 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 17284 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.0882, + "step": 17285 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 17286 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 17287 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 17288 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 17289 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 17290 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 17291 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 17292 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1828, + "step": 17293 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1707, + "step": 17294 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 17295 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 17296 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 17297 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 17298 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 17299 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 17300 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1706, + "step": 17301 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1526, + "step": 17302 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 17303 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 17304 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 17305 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 17306 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 17307 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 17308 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 17309 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 17310 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1495, + "step": 17311 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 17312 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 17313 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 17314 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 17315 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 17316 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 17317 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 17318 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 17319 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 17320 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 17321 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1478, + "step": 17322 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 17323 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 17324 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 17325 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1707, + "step": 17326 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 17327 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 17328 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1562, + "step": 17329 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 17330 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 17331 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 17332 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1812, + "step": 17333 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1878, + "step": 17334 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 17335 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 17336 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1642, + "step": 17337 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 17338 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1531, + "step": 17339 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 17340 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 17341 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 17342 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1867, + "step": 17343 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 17344 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 17345 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 17346 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 17347 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 17348 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 17349 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 17350 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 17351 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1676, + "step": 17352 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 17353 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.164, + "step": 17354 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 17355 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1726, + "step": 17356 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 17357 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 17358 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 17359 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 17360 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 17361 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 17362 + }, + { + "epoch": 1.05, + "learning_rate": 2e-05, + "loss": 0.1608, + "step": 17363 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 17364 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 17365 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1478, + "step": 17366 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 17367 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 17368 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 17369 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1488, + "step": 17370 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.0844, + "step": 17371 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 17372 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 17373 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 17374 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 17375 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1802, + "step": 17376 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 17377 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 17378 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 17379 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 17380 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 17381 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 17382 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 17383 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 17384 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 17385 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 17386 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 17387 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 17388 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 17389 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1476, + "step": 17390 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 17391 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 17392 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 17393 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 17394 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1674, + "step": 17395 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1503, + "step": 17396 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1768, + "step": 17397 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.0935, + "step": 17398 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 17399 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1561, + "step": 17400 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1595, + "step": 17401 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 17402 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 17403 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 17404 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 17405 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 17406 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 17407 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1541, + "step": 17408 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 17409 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1555, + "step": 17410 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 17411 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 17412 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 17413 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 17414 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 17415 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 17416 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 17417 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 17418 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 17419 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 17420 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 17421 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 17422 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 17423 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 17424 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 17425 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 17426 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 17427 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1627, + "step": 17428 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1855, + "step": 17429 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 17430 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 17431 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 17432 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 17433 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 17434 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 17435 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 17436 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1616, + "step": 17437 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 17438 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.0917, + "step": 17439 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.144, + "step": 17440 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1721, + "step": 17441 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 17442 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 17443 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 17444 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 17445 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1533, + "step": 17446 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 17447 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1617, + "step": 17448 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 17449 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 17450 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 17451 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 17452 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 17453 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 17454 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 17455 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1629, + "step": 17456 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 17457 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 17458 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 17459 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 17460 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1727, + "step": 17461 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1507, + "step": 17462 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 17463 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1524, + "step": 17464 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 17465 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 17466 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 17467 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 17468 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 17469 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 17470 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 17471 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 17472 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1702, + "step": 17473 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 17474 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 17475 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 17476 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 17477 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 17478 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.159, + "step": 17479 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.16, + "step": 17480 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1957, + "step": 17481 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1511, + "step": 17482 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 17483 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 17484 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 17485 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 17486 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1757, + "step": 17487 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1516, + "step": 17488 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1554, + "step": 17489 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1647, + "step": 17490 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 17491 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.145, + "step": 17492 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1684, + "step": 17493 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 17494 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.159, + "step": 17495 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 17496 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 17497 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 17498 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 17499 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 17500 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 17501 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 17502 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1585, + "step": 17503 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 17504 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 17505 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1549, + "step": 17506 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 17507 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 17508 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 17509 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 17510 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 17511 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1443, + "step": 17512 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 17513 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 17514 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 17515 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1972, + "step": 17516 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.145, + "step": 17517 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 17518 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 17519 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 17520 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 17521 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 17522 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1607, + "step": 17523 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1652, + "step": 17524 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 17525 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 17526 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.182, + "step": 17527 + }, + { + "epoch": 1.06, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 17528 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 17529 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1476, + "step": 17530 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1707, + "step": 17531 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 17532 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 17533 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 17534 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 17535 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 17536 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 17537 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 17538 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 17539 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 17540 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 17541 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 17542 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 17543 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 17544 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 17545 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 17546 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 17547 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 17548 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 17549 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 17550 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 17551 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 17552 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 17553 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 17554 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 17555 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 17556 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 17557 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 17558 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 17559 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 17560 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 17561 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 17562 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 17563 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1635, + "step": 17564 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 17565 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 17566 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 17567 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 17568 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 17569 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 17570 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 17571 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 17572 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 17573 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 17574 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1611, + "step": 17575 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 17576 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 17577 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 17578 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 17579 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1518, + "step": 17580 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.196, + "step": 17581 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1521, + "step": 17582 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 17583 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 17584 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 17585 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1812, + "step": 17586 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 17587 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 17588 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 17589 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 17590 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 17591 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.0932, + "step": 17592 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 17593 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 17594 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 17595 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 17596 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1597, + "step": 17597 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 17598 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 17599 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 17600 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 17601 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1632, + "step": 17602 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 17603 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1579, + "step": 17604 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1573, + "step": 17605 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 17606 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 17607 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.0895, + "step": 17608 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 17609 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 17610 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1814, + "step": 17611 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 17612 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 17613 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 17614 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 17615 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 17616 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 17617 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1524, + "step": 17618 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 17619 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1524, + "step": 17620 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1524, + "step": 17621 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1502, + "step": 17622 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 17623 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1687, + "step": 17624 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1756, + "step": 17625 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 17626 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 17627 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 17628 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 17629 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 17630 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 17631 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 17632 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 17633 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 17634 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 17635 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 17636 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1488, + "step": 17637 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1845, + "step": 17638 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 17639 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1922, + "step": 17640 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1573, + "step": 17641 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 17642 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 17643 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 17644 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 17645 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 17646 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 17647 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 17648 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1772, + "step": 17649 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 17650 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 17651 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 17652 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1635, + "step": 17653 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 17654 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 17655 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.169, + "step": 17656 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 17657 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 17658 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 17659 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 17660 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 17661 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 17662 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 17663 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 17664 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1806, + "step": 17665 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 17666 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 17667 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 17668 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.146, + "step": 17669 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 17670 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 17671 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 17672 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.172, + "step": 17673 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 17674 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 17675 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 17676 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1459, + "step": 17677 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 17678 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 17679 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1634, + "step": 17680 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1564, + "step": 17681 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 17682 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 17683 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1634, + "step": 17684 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 17685 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 17686 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 17687 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1719, + "step": 17688 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 17689 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 17690 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.0746, + "step": 17691 + }, + { + "epoch": 1.07, + "learning_rate": 2e-05, + "loss": 0.164, + "step": 17692 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 17693 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1655, + "step": 17694 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1739, + "step": 17695 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1736, + "step": 17696 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.2712, + "step": 17697 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 17698 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1468, + "step": 17699 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1637, + "step": 17700 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 17701 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 17702 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 17703 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.2022, + "step": 17704 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 17705 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1521, + "step": 17706 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 17707 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 17708 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.146, + "step": 17709 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 17710 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 17711 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1605, + "step": 17712 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 17713 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1698, + "step": 17714 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 17715 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 17716 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 17717 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 17718 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 17719 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 17720 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 17721 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 17722 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 17723 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 17724 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 17725 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 17726 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 17727 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 17728 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 17729 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 17730 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 17731 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 17732 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 17733 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 17734 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 17735 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 17736 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 17737 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 17738 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 17739 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 17740 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1513, + "step": 17741 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 17742 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 17743 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.0943, + "step": 17744 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 17745 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 17746 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 17747 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 17748 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 17749 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 17750 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 17751 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1674, + "step": 17752 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1462, + "step": 17753 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 17754 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.2012, + "step": 17755 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.148, + "step": 17756 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 17757 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1497, + "step": 17758 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 17759 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 17760 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.0878, + "step": 17761 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 17762 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 17763 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 17764 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 17765 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 17766 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1578, + "step": 17767 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 17768 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 17769 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 17770 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.153, + "step": 17771 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 17772 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1664, + "step": 17773 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 17774 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 17775 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1596, + "step": 17776 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1499, + "step": 17777 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 17778 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.165, + "step": 17779 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1669, + "step": 17780 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 17781 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 17782 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 17783 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1567, + "step": 17784 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 17785 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 17786 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 17787 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 17788 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 17789 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 17790 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1752, + "step": 17791 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 17792 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 17793 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 17794 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1734, + "step": 17795 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 17796 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 17797 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 17798 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 17799 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 17800 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 17801 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 17802 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 17803 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 17804 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 17805 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 17806 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.0868, + "step": 17807 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 17808 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 17809 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 17810 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1507, + "step": 17811 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 17812 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 17813 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1557, + "step": 17814 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 17815 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 17816 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 17817 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 17818 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.0832, + "step": 17819 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 17820 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1669, + "step": 17821 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 17822 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 17823 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1505, + "step": 17824 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 17825 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 17826 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 17827 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1642, + "step": 17828 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 17829 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 17830 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 17831 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 17832 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1577, + "step": 17833 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 17834 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 17835 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 17836 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.18, + "step": 17837 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.0891, + "step": 17838 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 17839 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 17840 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 17841 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 17842 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 17843 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1731, + "step": 17844 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 17845 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1847, + "step": 17846 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 17847 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 17848 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 17849 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 17850 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 17851 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 17852 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 17853 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 17854 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 17855 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1798, + "step": 17856 + }, + { + "epoch": 1.08, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 17857 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 17858 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1591, + "step": 17859 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 17860 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 17861 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 17862 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1763, + "step": 17863 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1616, + "step": 17864 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 17865 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 17866 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 17867 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 17868 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1616, + "step": 17869 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 17870 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1703, + "step": 17871 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 17872 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 17873 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1819, + "step": 17874 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 17875 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 17876 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 17877 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 17878 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 17879 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 17880 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 17881 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 17882 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1516, + "step": 17883 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.153, + "step": 17884 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 17885 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 17886 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 17887 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 17888 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 17889 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 17890 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 17891 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 17892 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1478, + "step": 17893 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 17894 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 17895 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 17896 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 17897 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 17898 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 17899 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1573, + "step": 17900 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 17901 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 17902 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 17903 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 17904 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 17905 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1741, + "step": 17906 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 17907 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1528, + "step": 17908 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 17909 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 17910 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.2058, + "step": 17911 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.147, + "step": 17912 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 17913 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 17914 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.0846, + "step": 17915 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 17916 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1637, + "step": 17917 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1782, + "step": 17918 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1755, + "step": 17919 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 17920 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 17921 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 17922 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1668, + "step": 17923 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1453, + "step": 17924 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 17925 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1875, + "step": 17926 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 17927 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 17928 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 17929 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 17930 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1697, + "step": 17931 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1589, + "step": 17932 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 17933 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 17934 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 17935 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1553, + "step": 17936 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 17937 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 17938 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1626, + "step": 17939 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 17940 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 17941 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 17942 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1447, + "step": 17943 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 17944 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1462, + "step": 17945 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 17946 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 17947 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1642, + "step": 17948 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 17949 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 17950 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 17951 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 17952 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 17953 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 17954 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 17955 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 17956 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.145, + "step": 17957 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 17958 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 17959 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1861, + "step": 17960 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 17961 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1573, + "step": 17962 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 17963 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 17964 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1771, + "step": 17965 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 17966 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 17967 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1684, + "step": 17968 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 17969 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 17970 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 17971 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 17972 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 17973 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 17974 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 17975 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 17976 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 17977 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 17978 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 17979 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 17980 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 17981 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 17982 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.0943, + "step": 17983 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 17984 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 17985 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 17986 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1846, + "step": 17987 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 17988 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 17989 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 17990 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1733, + "step": 17991 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 17992 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1698, + "step": 17993 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 17994 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 17995 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 17996 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1684, + "step": 17997 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 17998 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 17999 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 18000 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.18, + "step": 18001 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.196, + "step": 18002 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 18003 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.146, + "step": 18004 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 18005 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 18006 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 18007 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 18008 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 18009 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 18010 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 18011 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 18012 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1567, + "step": 18013 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 18014 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 18015 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 18016 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1599, + "step": 18017 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1694, + "step": 18018 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 18019 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1478, + "step": 18020 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 18021 + }, + { + "epoch": 1.09, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 18022 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 18023 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 18024 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 18025 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1672, + "step": 18026 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 18027 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 18028 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 18029 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1592, + "step": 18030 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 18031 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 18032 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 18033 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 18034 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 18035 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 18036 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 18037 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 18038 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 18039 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 18040 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1447, + "step": 18041 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 18042 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 18043 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 18044 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 18045 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 18046 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 18047 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.147, + "step": 18048 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 18049 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 18050 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 18051 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 18052 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 18053 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 18054 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.144, + "step": 18055 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1469, + "step": 18056 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 18057 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 18058 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1668, + "step": 18059 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 18060 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 18061 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 18062 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 18063 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1822, + "step": 18064 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 18065 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 18066 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 18067 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 18068 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 18069 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 18070 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 18071 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 18072 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.163, + "step": 18073 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 18074 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 18075 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 18076 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 18077 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1772, + "step": 18078 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 18079 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.158, + "step": 18080 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 18081 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.148, + "step": 18082 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1519, + "step": 18083 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 18084 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 18085 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 18086 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 18087 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 18088 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 18089 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 18090 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.155, + "step": 18091 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 18092 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 18093 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1951, + "step": 18094 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1613, + "step": 18095 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 18096 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.145, + "step": 18097 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1534, + "step": 18098 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 18099 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1532, + "step": 18100 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 18101 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1513, + "step": 18102 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1468, + "step": 18103 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1649, + "step": 18104 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 18105 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1815, + "step": 18106 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1605, + "step": 18107 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1648, + "step": 18108 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1478, + "step": 18109 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 18110 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 18111 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1515, + "step": 18112 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 18113 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 18114 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 18115 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 18116 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 18117 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 18118 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 18119 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1546, + "step": 18120 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.2068, + "step": 18121 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 18122 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1566, + "step": 18123 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 18124 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 18125 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 18126 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 18127 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 18128 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 18129 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1685, + "step": 18130 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 18131 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 18132 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 18133 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 18134 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1557, + "step": 18135 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 18136 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 18137 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 18138 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 18139 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 18140 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 18141 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 18142 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 18143 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 18144 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1679, + "step": 18145 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1586, + "step": 18146 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 18147 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 18148 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 18149 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 18150 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 18151 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1725, + "step": 18152 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 18153 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 18154 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1668, + "step": 18155 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 18156 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 18157 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 18158 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 18159 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 18160 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 18161 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1499, + "step": 18162 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 18163 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 18164 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1746, + "step": 18165 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 18166 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 18167 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 18168 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.148, + "step": 18169 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 18170 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 18171 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 18172 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 18173 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 18174 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 18175 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 18176 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1476, + "step": 18177 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 18178 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 18179 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 18180 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 18181 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1555, + "step": 18182 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 18183 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 18184 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1567, + "step": 18185 + }, + { + "epoch": 1.1, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 18186 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 18187 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 18188 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1541, + "step": 18189 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 18190 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 18191 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.0826, + "step": 18192 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 18193 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 18194 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 18195 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 18196 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1765, + "step": 18197 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 18198 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 18199 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 18200 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 18201 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 18202 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.2104, + "step": 18203 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 18204 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 18205 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 18206 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 18207 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 18208 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 18209 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1442, + "step": 18210 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1531, + "step": 18211 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1875, + "step": 18212 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.175, + "step": 18213 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1658, + "step": 18214 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 18215 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 18216 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 18217 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 18218 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 18219 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 18220 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1592, + "step": 18221 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 18222 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1628, + "step": 18223 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 18224 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 18225 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 18226 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 18227 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 18228 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1622, + "step": 18229 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 18230 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 18231 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1599, + "step": 18232 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1847, + "step": 18233 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 18234 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.158, + "step": 18235 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 18236 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 18237 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 18238 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 18239 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 18240 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 18241 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.165, + "step": 18242 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 18243 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 18244 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 18245 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1528, + "step": 18246 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 18247 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 18248 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1751, + "step": 18249 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 18250 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1546, + "step": 18251 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 18252 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 18253 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1772, + "step": 18254 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 18255 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 18256 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 18257 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 18258 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 18259 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.0891, + "step": 18260 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 18261 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1606, + "step": 18262 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1591, + "step": 18263 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 18264 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.191, + "step": 18265 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 18266 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1654, + "step": 18267 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1695, + "step": 18268 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 18269 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 18270 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1621, + "step": 18271 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 18272 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 18273 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 18274 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1734, + "step": 18275 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1982, + "step": 18276 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 18277 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 18278 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 18279 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 18280 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1801, + "step": 18281 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1712, + "step": 18282 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 18283 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 18284 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 18285 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 18286 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 18287 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 18288 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 18289 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 18290 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 18291 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1511, + "step": 18292 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 18293 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 18294 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1617, + "step": 18295 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 18296 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 18297 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 18298 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 18299 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 18300 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 18301 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1605, + "step": 18302 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 18303 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 18304 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1771, + "step": 18305 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 18306 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 18307 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 18308 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 18309 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1623, + "step": 18310 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 18311 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 18312 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 18313 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 18314 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 18315 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 18316 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 18317 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1662, + "step": 18318 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 18319 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 18320 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 18321 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.159, + "step": 18322 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 18323 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 18324 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 18325 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 18326 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 18327 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 18328 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 18329 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1456, + "step": 18330 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 18331 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 18332 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 18333 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 18334 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 18335 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 18336 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 18337 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 18338 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 18339 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1626, + "step": 18340 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 18341 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 18342 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1785, + "step": 18343 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 18344 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 18345 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1808, + "step": 18346 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 18347 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 18348 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1637, + "step": 18349 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 18350 + }, + { + "epoch": 1.11, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 18351 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1548, + "step": 18352 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 18353 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 18354 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1568, + "step": 18355 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 18356 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 18357 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1476, + "step": 18358 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 18359 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 18360 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1598, + "step": 18361 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 18362 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 18363 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 18364 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 18365 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 18366 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 18367 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 18368 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 18369 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 18370 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1653, + "step": 18371 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 18372 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 18373 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 18374 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 18375 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 18376 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 18377 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1468, + "step": 18378 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1671, + "step": 18379 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 18380 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 18381 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1778, + "step": 18382 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1453, + "step": 18383 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1785, + "step": 18384 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 18385 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 18386 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 18387 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 18388 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 18389 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 18390 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 18391 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 18392 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 18393 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 18394 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1634, + "step": 18395 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1526, + "step": 18396 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 18397 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 18398 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 18399 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 18400 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 18401 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 18402 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 18403 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 18404 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 18405 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1804, + "step": 18406 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 18407 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1626, + "step": 18408 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1663, + "step": 18409 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 18410 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 18411 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 18412 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 18413 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 18414 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.158, + "step": 18415 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1635, + "step": 18416 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1953, + "step": 18417 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 18418 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 18419 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 18420 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 18421 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 18422 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 18423 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 18424 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 18425 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1562, + "step": 18426 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 18427 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 18428 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 18429 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1515, + "step": 18430 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 18431 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1468, + "step": 18432 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 18433 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 18434 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 18435 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 18436 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 18437 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 18438 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 18439 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1579, + "step": 18440 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1667, + "step": 18441 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1511, + "step": 18442 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1629, + "step": 18443 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 18444 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1533, + "step": 18445 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 18446 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 18447 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 18448 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 18449 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 18450 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 18451 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 18452 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 18453 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1601, + "step": 18454 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.0872, + "step": 18455 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1791, + "step": 18456 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1621, + "step": 18457 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 18458 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.175, + "step": 18459 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1658, + "step": 18460 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 18461 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1498, + "step": 18462 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1628, + "step": 18463 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 18464 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 18465 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 18466 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 18467 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.2016, + "step": 18468 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 18469 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 18470 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 18471 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.156, + "step": 18472 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 18473 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 18474 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 18475 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 18476 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1583, + "step": 18477 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1788, + "step": 18478 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 18479 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 18480 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 18481 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1726, + "step": 18482 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 18483 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 18484 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 18485 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 18486 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 18487 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.0756, + "step": 18488 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 18489 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 18490 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 18491 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 18492 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 18493 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1747, + "step": 18494 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 18495 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 18496 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 18497 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 18498 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 18499 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 18500 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 18501 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 18502 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 18503 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 18504 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1761, + "step": 18505 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 18506 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 18507 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 18508 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 18509 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 18510 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 18511 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1803, + "step": 18512 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 18513 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.1554, + "step": 18514 + }, + { + "epoch": 1.12, + "learning_rate": 2e-05, + "loss": 0.146, + "step": 18515 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1459, + "step": 18516 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 18517 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1526, + "step": 18518 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 18519 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 18520 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 18521 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 18522 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 18523 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 18524 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 18525 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 18526 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1714, + "step": 18527 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.2128, + "step": 18528 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1602, + "step": 18529 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 18530 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 18531 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.173, + "step": 18532 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 18533 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1971, + "step": 18534 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 18535 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1622, + "step": 18536 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1503, + "step": 18537 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 18538 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 18539 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1468, + "step": 18540 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 18541 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 18542 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 18543 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1451, + "step": 18544 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 18545 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 18546 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 18547 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1443, + "step": 18548 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 18549 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 18550 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 18551 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 18552 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 18553 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 18554 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 18555 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1534, + "step": 18556 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 18557 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 18558 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 18559 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 18560 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.2114, + "step": 18561 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 18562 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1723, + "step": 18563 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 18564 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 18565 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1607, + "step": 18566 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 18567 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 18568 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 18569 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 18570 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 18571 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1618, + "step": 18572 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1443, + "step": 18573 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 18574 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 18575 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 18576 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.148, + "step": 18577 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 18578 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 18579 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 18580 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1677, + "step": 18581 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 18582 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 18583 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.214, + "step": 18584 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 18585 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 18586 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 18587 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 18588 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1535, + "step": 18589 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1534, + "step": 18590 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.2062, + "step": 18591 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1503, + "step": 18592 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 18593 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1497, + "step": 18594 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 18595 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 18596 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1499, + "step": 18597 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 18598 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 18599 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 18600 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1786, + "step": 18601 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 18602 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 18603 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 18604 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 18605 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 18606 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1656, + "step": 18607 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1553, + "step": 18608 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 18609 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1721, + "step": 18610 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1502, + "step": 18611 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 18612 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 18613 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 18614 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 18615 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.153, + "step": 18616 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1637, + "step": 18617 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 18618 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 18619 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 18620 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 18621 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 18622 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 18623 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 18624 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 18625 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 18626 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 18627 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 18628 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 18629 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 18630 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1631, + "step": 18631 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 18632 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.147, + "step": 18633 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 18634 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.17, + "step": 18635 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 18636 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 18637 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 18638 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1582, + "step": 18639 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 18640 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 18641 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 18642 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 18643 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 18644 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 18645 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 18646 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 18647 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 18648 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 18649 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1546, + "step": 18650 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1746, + "step": 18651 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 18652 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 18653 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 18654 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 18655 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 18656 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 18657 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 18658 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 18659 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 18660 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 18661 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 18662 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 18663 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 18664 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 18665 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 18666 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1664, + "step": 18667 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 18668 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 18669 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 18670 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 18671 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.0911, + "step": 18672 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 18673 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 18674 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 18675 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 18676 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.159, + "step": 18677 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 18678 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1921, + "step": 18679 + }, + { + "epoch": 1.13, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 18680 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 18681 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 18682 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 18683 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1655, + "step": 18684 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 18685 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 18686 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 18687 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 18688 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1686, + "step": 18689 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 18690 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1516, + "step": 18691 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 18692 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 18693 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 18694 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 18695 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 18696 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 18697 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 18698 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 18699 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 18700 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 18701 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 18702 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 18703 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1599, + "step": 18704 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 18705 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 18706 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 18707 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 18708 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1591, + "step": 18709 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 18710 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 18711 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 18712 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 18713 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 18714 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 18715 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 18716 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 18717 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 18718 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 18719 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.0836, + "step": 18720 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 18721 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 18722 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 18723 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 18724 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 18725 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 18726 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.0846, + "step": 18727 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 18728 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 18729 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1429, + "step": 18730 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 18731 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 18732 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 18733 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 18734 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.144, + "step": 18735 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1803, + "step": 18736 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 18737 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.0843, + "step": 18738 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1628, + "step": 18739 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 18740 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 18741 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 18742 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 18743 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1634, + "step": 18744 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1516, + "step": 18745 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 18746 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.178, + "step": 18747 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 18748 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 18749 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 18750 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 18751 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1743, + "step": 18752 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1653, + "step": 18753 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 18754 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1833, + "step": 18755 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 18756 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 18757 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1517, + "step": 18758 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 18759 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1532, + "step": 18760 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 18761 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 18762 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 18763 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.146, + "step": 18764 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 18765 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1638, + "step": 18766 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 18767 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 18768 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 18769 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 18770 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 18771 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 18772 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 18773 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 18774 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 18775 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 18776 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 18777 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.162, + "step": 18778 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 18779 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 18780 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1624, + "step": 18781 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 18782 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1505, + "step": 18783 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 18784 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 18785 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 18786 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 18787 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1499, + "step": 18788 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1758, + "step": 18789 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.2091, + "step": 18790 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1707, + "step": 18791 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 18792 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 18793 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 18794 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 18795 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 18796 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1505, + "step": 18797 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1716, + "step": 18798 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 18799 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 18800 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 18801 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 18802 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1505, + "step": 18803 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 18804 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 18805 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 18806 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1643, + "step": 18807 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 18808 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 18809 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 18810 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1797, + "step": 18811 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1617, + "step": 18812 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 18813 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1716, + "step": 18814 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 18815 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 18816 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1553, + "step": 18817 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 18818 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 18819 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 18820 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 18821 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 18822 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 18823 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 18824 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1704, + "step": 18825 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 18826 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 18827 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 18828 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 18829 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.0967, + "step": 18830 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 18831 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 18832 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 18833 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 18834 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1517, + "step": 18835 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 18836 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 18837 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 18838 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 18839 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 18840 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 18841 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 18842 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1664, + "step": 18843 + }, + { + "epoch": 1.14, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 18844 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 18845 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 18846 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1637, + "step": 18847 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 18848 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 18849 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 18850 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.153, + "step": 18851 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 18852 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 18853 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 18854 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 18855 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.16, + "step": 18856 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 18857 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.174, + "step": 18858 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 18859 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 18860 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1476, + "step": 18861 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1664, + "step": 18862 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 18863 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 18864 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.0941, + "step": 18865 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 18866 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 18867 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1643, + "step": 18868 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 18869 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 18870 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 18871 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 18872 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 18873 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 18874 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.0775, + "step": 18875 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 18876 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 18877 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1929, + "step": 18878 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 18879 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 18880 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 18881 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 18882 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 18883 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 18884 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1637, + "step": 18885 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 18886 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 18887 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 18888 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 18889 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 18890 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1647, + "step": 18891 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 18892 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1586, + "step": 18893 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 18894 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 18895 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 18896 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 18897 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1564, + "step": 18898 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 18899 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 18900 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 18901 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 18902 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1673, + "step": 18903 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1658, + "step": 18904 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 18905 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 18906 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 18907 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1635, + "step": 18908 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 18909 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 18910 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 18911 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 18912 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 18913 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 18914 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 18915 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 18916 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1576, + "step": 18917 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 18918 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 18919 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 18920 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 18921 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 18922 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 18923 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 18924 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 18925 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 18926 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1531, + "step": 18927 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 18928 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 18929 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 18930 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1563, + "step": 18931 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 18932 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1779, + "step": 18933 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1692, + "step": 18934 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 18935 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 18936 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 18937 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 18938 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 18939 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 18940 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 18941 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 18942 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 18943 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 18944 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1809, + "step": 18945 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 18946 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 18947 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1607, + "step": 18948 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 18949 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 18950 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1593, + "step": 18951 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1772, + "step": 18952 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 18953 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1524, + "step": 18954 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 18955 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 18956 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 18957 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 18958 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 18959 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 18960 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 18961 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 18962 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1614, + "step": 18963 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 18964 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 18965 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 18966 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 18967 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 18968 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 18969 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 18970 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 18971 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 18972 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 18973 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 18974 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1602, + "step": 18975 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 18976 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 18977 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 18978 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1646, + "step": 18979 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 18980 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 18981 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 18982 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1622, + "step": 18983 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 18984 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 18985 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1621, + "step": 18986 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 18987 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.161, + "step": 18988 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 18989 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 18990 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 18991 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 18992 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1836, + "step": 18993 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1592, + "step": 18994 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 18995 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 18996 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1608, + "step": 18997 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 18998 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1573, + "step": 18999 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 19000 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 19001 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 19002 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 19003 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1789, + "step": 19004 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1803, + "step": 19005 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 19006 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.0868, + "step": 19007 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 19008 + }, + { + "epoch": 1.15, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 19009 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 19010 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 19011 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 19012 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 19013 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 19014 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 19015 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 19016 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.177, + "step": 19017 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 19018 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 19019 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 19020 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.0837, + "step": 19021 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 19022 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 19023 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 19024 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 19025 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 19026 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 19027 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 19028 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 19029 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 19030 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 19031 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1553, + "step": 19032 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 19033 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 19034 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 19035 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 19036 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 19037 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 19038 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 19039 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1892, + "step": 19040 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 19041 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 19042 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 19043 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 19044 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.159, + "step": 19045 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 19046 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 19047 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1688, + "step": 19048 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 19049 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 19050 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 19051 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 19052 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 19053 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 19054 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1591, + "step": 19055 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.0898, + "step": 19056 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 19057 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1767, + "step": 19058 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1462, + "step": 19059 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 19060 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 19061 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 19062 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 19063 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 19064 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 19065 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 19066 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 19067 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1541, + "step": 19068 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 19069 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 19070 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 19071 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 19072 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.2031, + "step": 19073 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1553, + "step": 19074 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 19075 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 19076 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1613, + "step": 19077 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 19078 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1833, + "step": 19079 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.156, + "step": 19080 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 19081 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1499, + "step": 19082 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1511, + "step": 19083 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1553, + "step": 19084 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1737, + "step": 19085 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 19086 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 19087 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 19088 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 19089 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 19090 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 19091 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 19092 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.2051, + "step": 19093 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 19094 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1687, + "step": 19095 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 19096 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 19097 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 19098 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 19099 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1597, + "step": 19100 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 19101 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1585, + "step": 19102 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 19103 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 19104 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 19105 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 19106 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 19107 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 19108 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 19109 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 19110 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 19111 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 19112 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.155, + "step": 19113 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 19114 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 19115 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 19116 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 19117 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 19118 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 19119 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 19120 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1703, + "step": 19121 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 19122 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 19123 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1745, + "step": 19124 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 19125 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 19126 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 19127 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 19128 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 19129 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 19130 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1625, + "step": 19131 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 19132 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 19133 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 19134 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 19135 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 19136 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 19137 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 19138 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 19139 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 19140 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1991, + "step": 19141 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 19142 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 19143 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 19144 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 19145 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1528, + "step": 19146 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.169, + "step": 19147 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 19148 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 19149 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 19150 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 19151 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 19152 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.148, + "step": 19153 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 19154 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 19155 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 19156 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 19157 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 19158 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 19159 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 19160 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1554, + "step": 19161 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 19162 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1784, + "step": 19163 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 19164 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 19165 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 19166 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 19167 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1654, + "step": 19168 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 19169 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.161, + "step": 19170 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1817, + "step": 19171 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 19172 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 19173 + }, + { + "epoch": 1.16, + "learning_rate": 2e-05, + "loss": 0.1627, + "step": 19174 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 19175 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 19176 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 19177 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 19178 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1532, + "step": 19179 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 19180 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 19181 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 19182 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 19183 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 19184 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 19185 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 19186 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 19187 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 19188 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1551, + "step": 19189 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1521, + "step": 19190 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1671, + "step": 19191 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.169, + "step": 19192 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.153, + "step": 19193 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 19194 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.2002, + "step": 19195 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 19196 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 19197 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 19198 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 19199 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 19200 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 19201 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 19202 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 19203 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 19204 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 19205 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 19206 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1917, + "step": 19207 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 19208 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 19209 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 19210 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 19211 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 19212 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 19213 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 19214 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 19215 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 19216 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 19217 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 19218 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 19219 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 19220 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 19221 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 19222 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 19223 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 19224 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 19225 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 19226 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 19227 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 19228 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1728, + "step": 19229 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 19230 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 19231 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 19232 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 19233 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1603, + "step": 19234 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 19235 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1654, + "step": 19236 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 19237 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 19238 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1526, + "step": 19239 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 19240 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 19241 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1771, + "step": 19242 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 19243 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 19244 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 19245 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 19246 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1573, + "step": 19247 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1938, + "step": 19248 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 19249 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 19250 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 19251 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1465, + "step": 19252 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 19253 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 19254 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 19255 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 19256 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 19257 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1873, + "step": 19258 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 19259 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 19260 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 19261 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 19262 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1469, + "step": 19263 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1625, + "step": 19264 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 19265 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 19266 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1495, + "step": 19267 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 19268 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 19269 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 19270 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 19271 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 19272 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 19273 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 19274 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 19275 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1567, + "step": 19276 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 19277 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 19278 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 19279 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 19280 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 19281 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 19282 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1585, + "step": 19283 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 19284 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 19285 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 19286 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 19287 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1692, + "step": 19288 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1596, + "step": 19289 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.145, + "step": 19290 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1651, + "step": 19291 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 19292 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1534, + "step": 19293 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1767, + "step": 19294 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 19295 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1488, + "step": 19296 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 19297 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 19298 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1759, + "step": 19299 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 19300 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 19301 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1841, + "step": 19302 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 19303 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 19304 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 19305 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 19306 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 19307 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 19308 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 19309 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1793, + "step": 19310 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 19311 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 19312 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 19313 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 19314 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 19315 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 19316 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 19317 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 19318 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 19319 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 19320 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 19321 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 19322 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 19323 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 19324 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 19325 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 19326 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 19327 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1692, + "step": 19328 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 19329 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1666, + "step": 19330 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1766, + "step": 19331 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 19332 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 19333 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 19334 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1706, + "step": 19335 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 19336 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 19337 + }, + { + "epoch": 1.17, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 19338 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 19339 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 19340 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1465, + "step": 19341 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 19342 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 19343 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 19344 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.167, + "step": 19345 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 19346 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 19347 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 19348 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 19349 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 19350 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 19351 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 19352 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 19353 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 19354 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 19355 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 19356 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1476, + "step": 19357 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1542, + "step": 19358 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1601, + "step": 19359 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1574, + "step": 19360 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 19361 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 19362 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 19363 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.171, + "step": 19364 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 19365 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 19366 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 19367 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 19368 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 19369 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.205, + "step": 19370 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1719, + "step": 19371 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 19372 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1525, + "step": 19373 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 19374 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 19375 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1558, + "step": 19376 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 19377 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 19378 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 19379 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 19380 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1762, + "step": 19381 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 19382 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1623, + "step": 19383 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 19384 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1789, + "step": 19385 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 19386 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1948, + "step": 19387 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 19388 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1521, + "step": 19389 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 19390 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 19391 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 19392 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1668, + "step": 19393 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1653, + "step": 19394 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1447, + "step": 19395 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1659, + "step": 19396 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 19397 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1588, + "step": 19398 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 19399 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1874, + "step": 19400 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 19401 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1677, + "step": 19402 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 19403 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 19404 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 19405 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 19406 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 19407 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 19408 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1701, + "step": 19409 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 19410 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 19411 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1678, + "step": 19412 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 19413 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 19414 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 19415 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1746, + "step": 19416 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1574, + "step": 19417 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1833, + "step": 19418 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 19419 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 19420 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1852, + "step": 19421 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 19422 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 19423 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 19424 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 19425 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1598, + "step": 19426 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1768, + "step": 19427 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 19428 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.2229, + "step": 19429 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 19430 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 19431 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1499, + "step": 19432 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 19433 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 19434 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1614, + "step": 19435 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.2057, + "step": 19436 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 19437 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 19438 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 19439 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 19440 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1663, + "step": 19441 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 19442 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1727, + "step": 19443 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 19444 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1642, + "step": 19445 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 19446 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 19447 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 19448 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 19449 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 19450 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 19451 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 19452 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1649, + "step": 19453 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1518, + "step": 19454 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 19455 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 19456 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 19457 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 19458 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 19459 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 19460 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 19461 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 19462 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 19463 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1532, + "step": 19464 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 19465 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 19466 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 19467 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 19468 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1721, + "step": 19469 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 19470 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 19471 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.191, + "step": 19472 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 19473 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 19474 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 19475 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 19476 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 19477 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 19478 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.159, + "step": 19479 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 19480 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1586, + "step": 19481 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 19482 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 19483 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 19484 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 19485 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 19486 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 19487 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 19488 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 19489 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1502, + "step": 19490 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 19491 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 19492 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 19493 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 19494 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 19495 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 19496 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 19497 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 19498 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 19499 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 19500 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 19501 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 19502 + }, + { + "epoch": 1.18, + "learning_rate": 2e-05, + "loss": 0.1701, + "step": 19503 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1739, + "step": 19504 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 19505 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 19506 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 19507 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 19508 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1485, + "step": 19509 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1797, + "step": 19510 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 19511 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 19512 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1532, + "step": 19513 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 19514 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 19515 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 19516 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 19517 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1515, + "step": 19518 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 19519 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 19520 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 19521 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 19522 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 19523 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 19524 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.145, + "step": 19525 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 19526 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 19527 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 19528 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1578, + "step": 19529 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 19530 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1675, + "step": 19531 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 19532 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 19533 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 19534 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 19535 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 19536 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.2186, + "step": 19537 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 19538 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 19539 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 19540 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1768, + "step": 19541 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 19542 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 19543 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 19544 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 19545 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 19546 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 19547 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1593, + "step": 19548 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 19549 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 19550 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 19551 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.162, + "step": 19552 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 19553 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 19554 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1852, + "step": 19555 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 19556 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 19557 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1469, + "step": 19558 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 19559 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 19560 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 19561 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 19562 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 19563 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 19564 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 19565 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 19566 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 19567 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 19568 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1602, + "step": 19569 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 19570 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 19571 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1485, + "step": 19572 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 19573 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1589, + "step": 19574 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 19575 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.183, + "step": 19576 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1505, + "step": 19577 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 19578 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.161, + "step": 19579 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.0918, + "step": 19580 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 19581 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 19582 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 19583 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 19584 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 19585 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 19586 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 19587 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 19588 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 19589 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 19590 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 19591 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 19592 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 19593 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 19594 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1459, + "step": 19595 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 19596 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 19597 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 19598 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1794, + "step": 19599 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 19600 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 19601 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1798, + "step": 19602 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 19603 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.164, + "step": 19604 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 19605 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 19606 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 19607 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 19608 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 19609 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1534, + "step": 19610 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 19611 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.164, + "step": 19612 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 19613 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 19614 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 19615 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 19616 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1476, + "step": 19617 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 19618 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 19619 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1524, + "step": 19620 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 19621 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 19622 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 19623 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1502, + "step": 19624 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1605, + "step": 19625 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 19626 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 19627 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 19628 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1554, + "step": 19629 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 19630 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 19631 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 19632 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1485, + "step": 19633 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 19634 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 19635 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 19636 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 19637 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 19638 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 19639 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 19640 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1731, + "step": 19641 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 19642 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 19643 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 19644 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 19645 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 19646 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 19647 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1983, + "step": 19648 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 19649 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 19650 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 19651 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 19652 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 19653 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 19654 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 19655 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 19656 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 19657 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 19658 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1794, + "step": 19659 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 19660 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 19661 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 19662 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1617, + "step": 19663 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 19664 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 19665 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 19666 + }, + { + "epoch": 1.19, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 19667 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 19668 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 19669 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1853, + "step": 19670 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1558, + "step": 19671 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 19672 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1517, + "step": 19673 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1674, + "step": 19674 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 19675 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 19676 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 19677 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 19678 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1844, + "step": 19679 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 19680 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 19681 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 19682 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 19683 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.177, + "step": 19684 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 19685 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1567, + "step": 19686 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 19687 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 19688 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1574, + "step": 19689 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1557, + "step": 19690 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 19691 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 19692 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 19693 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1499, + "step": 19694 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 19695 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 19696 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 19697 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 19698 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 19699 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 19700 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 19701 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 19702 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 19703 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.0897, + "step": 19704 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 19705 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1505, + "step": 19706 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 19707 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 19708 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 19709 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 19710 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 19711 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1593, + "step": 19712 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 19713 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 19714 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 19715 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1574, + "step": 19716 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 19717 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 19718 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 19719 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 19720 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 19721 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 19722 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 19723 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 19724 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 19725 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 19726 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 19727 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 19728 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1626, + "step": 19729 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1822, + "step": 19730 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 19731 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 19732 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 19733 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 19734 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 19735 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 19736 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 19737 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 19738 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 19739 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1732, + "step": 19740 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1462, + "step": 19741 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 19742 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 19743 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 19744 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 19745 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 19746 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 19747 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 19748 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 19749 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 19750 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 19751 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1604, + "step": 19752 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1447, + "step": 19753 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 19754 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1462, + "step": 19755 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1635, + "step": 19756 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 19757 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 19758 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 19759 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 19760 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 19761 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 19762 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 19763 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1745, + "step": 19764 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 19765 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 19766 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1498, + "step": 19767 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 19768 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 19769 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1518, + "step": 19770 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 19771 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 19772 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 19773 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 19774 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 19775 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1526, + "step": 19776 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 19777 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 19778 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 19779 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 19780 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1847, + "step": 19781 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1787, + "step": 19782 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1626, + "step": 19783 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 19784 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 19785 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.152, + "step": 19786 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.158, + "step": 19787 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1701, + "step": 19788 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1706, + "step": 19789 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 19790 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 19791 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1515, + "step": 19792 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 19793 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 19794 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1552, + "step": 19795 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 19796 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 19797 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 19798 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1788, + "step": 19799 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 19800 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 19801 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1651, + "step": 19802 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 19803 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 19804 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1788, + "step": 19805 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1525, + "step": 19806 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 19807 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 19808 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1655, + "step": 19809 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1443, + "step": 19810 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 19811 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 19812 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 19813 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 19814 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.164, + "step": 19815 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1704, + "step": 19816 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.144, + "step": 19817 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 19818 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 19819 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 19820 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1555, + "step": 19821 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 19822 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1498, + "step": 19823 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 19824 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 19825 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 19826 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 19827 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1513, + "step": 19828 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 19829 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 19830 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1531, + "step": 19831 + }, + { + "epoch": 1.2, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 19832 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1447, + "step": 19833 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1511, + "step": 19834 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1622, + "step": 19835 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 19836 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 19837 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 19838 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 19839 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 19840 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 19841 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 19842 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 19843 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.0929, + "step": 19844 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 19845 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1562, + "step": 19846 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 19847 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1911, + "step": 19848 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 19849 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1598, + "step": 19850 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 19851 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1911, + "step": 19852 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 19853 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 19854 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 19855 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1534, + "step": 19856 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 19857 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 19858 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 19859 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1582, + "step": 19860 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 19861 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1627, + "step": 19862 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 19863 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1685, + "step": 19864 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1624, + "step": 19865 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 19866 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 19867 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1478, + "step": 19868 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1582, + "step": 19869 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 19870 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 19871 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 19872 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1429, + "step": 19873 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 19874 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 19875 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 19876 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 19877 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 19878 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 19879 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 19880 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 19881 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 19882 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 19883 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 19884 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 19885 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 19886 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 19887 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1833, + "step": 19888 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1782, + "step": 19889 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 19890 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1533, + "step": 19891 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1851, + "step": 19892 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1644, + "step": 19893 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 19894 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 19895 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 19896 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 19897 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 19898 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1647, + "step": 19899 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 19900 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 19901 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 19902 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 19903 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 19904 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 19905 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 19906 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.156, + "step": 19907 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 19908 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 19909 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1654, + "step": 19910 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 19911 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 19912 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 19913 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1574, + "step": 19914 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 19915 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 19916 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 19917 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 19918 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 19919 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 19920 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 19921 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 19922 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 19923 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 19924 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 19925 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 19926 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 19927 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1613, + "step": 19928 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 19929 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 19930 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 19931 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 19932 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 19933 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 19934 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1714, + "step": 19935 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 19936 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 19937 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 19938 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 19939 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1689, + "step": 19940 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1797, + "step": 19941 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 19942 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.2003, + "step": 19943 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1652, + "step": 19944 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 19945 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 19946 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 19947 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 19948 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 19949 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 19950 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 19951 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 19952 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1937, + "step": 19953 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 19954 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 19955 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1625, + "step": 19956 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 19957 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.169, + "step": 19958 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 19959 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 19960 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 19961 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 19962 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 19963 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 19964 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1709, + "step": 19965 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 19966 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.158, + "step": 19967 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1599, + "step": 19968 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.161, + "step": 19969 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 19970 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 19971 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 19972 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 19973 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 19974 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 19975 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1602, + "step": 19976 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 19977 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1925, + "step": 19978 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 19979 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 19980 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 19981 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 19982 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 19983 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 19984 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 19985 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1666, + "step": 19986 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1845, + "step": 19987 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 19988 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1585, + "step": 19989 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 19990 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 19991 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 19992 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 19993 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 19994 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 19995 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 19996 + }, + { + "epoch": 1.21, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 19997 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 19998 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 19999 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 20000 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 20001 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.147, + "step": 20002 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 20003 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 20004 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 20005 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 20006 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1923, + "step": 20007 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 20008 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 20009 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 20010 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 20011 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 20012 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1702, + "step": 20013 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 20014 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 20015 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 20016 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1689, + "step": 20017 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 20018 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 20019 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 20020 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1769, + "step": 20021 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 20022 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1503, + "step": 20023 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1824, + "step": 20024 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1535, + "step": 20025 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1871, + "step": 20026 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1585, + "step": 20027 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1949, + "step": 20028 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 20029 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1563, + "step": 20030 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 20031 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 20032 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 20033 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1713, + "step": 20034 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 20035 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 20036 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 20037 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 20038 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 20039 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1478, + "step": 20040 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 20041 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 20042 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 20043 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 20044 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 20045 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 20046 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 20047 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1665, + "step": 20048 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1562, + "step": 20049 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 20050 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 20051 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 20052 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 20053 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1552, + "step": 20054 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 20055 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1703, + "step": 20056 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1707, + "step": 20057 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 20058 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 20059 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 20060 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 20061 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 20062 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1893, + "step": 20063 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 20064 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 20065 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 20066 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1986, + "step": 20067 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 20068 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 20069 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1649, + "step": 20070 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 20071 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 20072 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 20073 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1779, + "step": 20074 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 20075 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 20076 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 20077 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.0943, + "step": 20078 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1521, + "step": 20079 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 20080 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 20081 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 20082 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1531, + "step": 20083 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 20084 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 20085 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 20086 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 20087 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1526, + "step": 20088 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 20089 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 20090 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.144, + "step": 20091 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 20092 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 20093 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 20094 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 20095 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 20096 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 20097 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1716, + "step": 20098 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 20099 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 20100 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 20101 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 20102 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1546, + "step": 20103 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 20104 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.152, + "step": 20105 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 20106 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 20107 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 20108 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 20109 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1582, + "step": 20110 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 20111 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 20112 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 20113 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 20114 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 20115 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 20116 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1627, + "step": 20117 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 20118 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 20119 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1485, + "step": 20120 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.153, + "step": 20121 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 20122 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 20123 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 20124 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1626, + "step": 20125 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 20126 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 20127 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 20128 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1495, + "step": 20129 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1564, + "step": 20130 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 20131 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 20132 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 20133 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 20134 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 20135 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 20136 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 20137 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 20138 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 20139 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1812, + "step": 20140 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1682, + "step": 20141 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 20142 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 20143 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 20144 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 20145 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 20146 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 20147 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 20148 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1517, + "step": 20149 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1747, + "step": 20150 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1639, + "step": 20151 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 20152 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 20153 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 20154 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 20155 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 20156 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 20157 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.162, + "step": 20158 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 20159 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 20160 + }, + { + "epoch": 1.22, + "learning_rate": 2e-05, + "loss": 0.172, + "step": 20161 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 20162 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 20163 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 20164 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 20165 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 20166 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 20167 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 20168 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 20169 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 20170 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 20171 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 20172 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 20173 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1497, + "step": 20174 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1459, + "step": 20175 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 20176 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1621, + "step": 20177 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 20178 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 20179 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 20180 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 20181 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1561, + "step": 20182 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1541, + "step": 20183 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 20184 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1624, + "step": 20185 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 20186 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 20187 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1688, + "step": 20188 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 20189 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 20190 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.147, + "step": 20191 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 20192 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1851, + "step": 20193 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 20194 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1729, + "step": 20195 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1826, + "step": 20196 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 20197 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 20198 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1553, + "step": 20199 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 20200 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 20201 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1524, + "step": 20202 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 20203 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.2025, + "step": 20204 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 20205 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1655, + "step": 20206 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 20207 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1499, + "step": 20208 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 20209 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 20210 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 20211 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 20212 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 20213 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 20214 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 20215 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1591, + "step": 20216 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 20217 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 20218 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.2151, + "step": 20219 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 20220 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1592, + "step": 20221 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1567, + "step": 20222 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.17, + "step": 20223 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 20224 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 20225 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 20226 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 20227 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 20228 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1664, + "step": 20229 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.17, + "step": 20230 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.0892, + "step": 20231 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 20232 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1679, + "step": 20233 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 20234 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1712, + "step": 20235 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 20236 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1574, + "step": 20237 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 20238 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 20239 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 20240 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 20241 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1614, + "step": 20242 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 20243 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 20244 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 20245 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1443, + "step": 20246 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 20247 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 20248 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1867, + "step": 20249 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 20250 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1686, + "step": 20251 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1845, + "step": 20252 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 20253 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 20254 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 20255 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 20256 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 20257 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 20258 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1576, + "step": 20259 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 20260 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.191, + "step": 20261 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 20262 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1505, + "step": 20263 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1601, + "step": 20264 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 20265 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 20266 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 20267 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 20268 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 20269 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 20270 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 20271 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 20272 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1667, + "step": 20273 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 20274 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 20275 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 20276 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 20277 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 20278 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 20279 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 20280 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1806, + "step": 20281 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1875, + "step": 20282 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1644, + "step": 20283 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 20284 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 20285 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 20286 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1754, + "step": 20287 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 20288 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 20289 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 20290 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 20291 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 20292 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1525, + "step": 20293 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 20294 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 20295 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 20296 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 20297 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 20298 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 20299 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1567, + "step": 20300 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 20301 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 20302 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 20303 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 20304 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 20305 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1567, + "step": 20306 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.144, + "step": 20307 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 20308 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 20309 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1911, + "step": 20310 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 20311 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 20312 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 20313 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 20314 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 20315 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 20316 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 20317 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.144, + "step": 20318 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1629, + "step": 20319 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 20320 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 20321 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 20322 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 20323 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 20324 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 20325 + }, + { + "epoch": 1.23, + "learning_rate": 2e-05, + "loss": 0.1558, + "step": 20326 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 20327 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1865, + "step": 20328 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1577, + "step": 20329 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 20330 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1564, + "step": 20331 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1526, + "step": 20332 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 20333 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 20334 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 20335 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 20336 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 20337 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1637, + "step": 20338 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 20339 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 20340 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 20341 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 20342 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1649, + "step": 20343 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 20344 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 20345 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 20346 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 20347 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 20348 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 20349 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 20350 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1526, + "step": 20351 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1608, + "step": 20352 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 20353 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 20354 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 20355 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 20356 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 20357 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 20358 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 20359 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 20360 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1507, + "step": 20361 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 20362 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 20363 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 20364 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 20365 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 20366 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 20367 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 20368 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 20369 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 20370 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1527, + "step": 20371 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1833, + "step": 20372 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 20373 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 20374 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 20375 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 20376 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 20377 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 20378 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 20379 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 20380 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 20381 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1527, + "step": 20382 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 20383 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 20384 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 20385 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 20386 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 20387 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 20388 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1733, + "step": 20389 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1765, + "step": 20390 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 20391 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 20392 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 20393 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 20394 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 20395 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 20396 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 20397 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 20398 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 20399 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 20400 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 20401 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 20402 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 20403 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 20404 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 20405 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1614, + "step": 20406 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 20407 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1527, + "step": 20408 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 20409 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 20410 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1686, + "step": 20411 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 20412 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 20413 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.163, + "step": 20414 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1604, + "step": 20415 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.178, + "step": 20416 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1708, + "step": 20417 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1511, + "step": 20418 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1734, + "step": 20419 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 20420 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 20421 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 20422 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 20423 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 20424 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 20425 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 20426 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 20427 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 20428 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 20429 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.09, + "step": 20430 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 20431 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 20432 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1844, + "step": 20433 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1488, + "step": 20434 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 20435 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 20436 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 20437 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1697, + "step": 20438 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 20439 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 20440 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 20441 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 20442 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1975, + "step": 20443 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 20444 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 20445 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 20446 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 20447 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 20448 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 20449 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 20450 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1616, + "step": 20451 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 20452 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.0901, + "step": 20453 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1716, + "step": 20454 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 20455 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 20456 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 20457 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 20458 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 20459 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1442, + "step": 20460 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1742, + "step": 20461 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 20462 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 20463 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 20464 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1981, + "step": 20465 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 20466 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.158, + "step": 20467 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 20468 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 20469 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 20470 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 20471 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 20472 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 20473 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 20474 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 20475 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 20476 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 20477 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1646, + "step": 20478 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1927, + "step": 20479 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1503, + "step": 20480 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 20481 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 20482 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 20483 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 20484 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.182, + "step": 20485 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 20486 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.0883, + "step": 20487 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 20488 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 20489 + }, + { + "epoch": 1.24, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 20490 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 20491 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 20492 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 20493 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 20494 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 20495 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1451, + "step": 20496 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 20497 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 20498 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1642, + "step": 20499 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 20500 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1965, + "step": 20501 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1816, + "step": 20502 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 20503 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1468, + "step": 20504 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 20505 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1817, + "step": 20506 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 20507 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1731, + "step": 20508 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 20509 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1558, + "step": 20510 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1747, + "step": 20511 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 20512 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 20513 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.0838, + "step": 20514 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 20515 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1694, + "step": 20516 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 20517 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 20518 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1667, + "step": 20519 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.168, + "step": 20520 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1592, + "step": 20521 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 20522 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1839, + "step": 20523 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 20524 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 20525 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1557, + "step": 20526 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1577, + "step": 20527 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 20528 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 20529 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1932, + "step": 20530 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 20531 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1706, + "step": 20532 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 20533 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 20534 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 20535 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 20536 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 20537 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 20538 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 20539 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 20540 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 20541 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 20542 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 20543 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 20544 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 20545 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 20546 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 20547 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 20548 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 20549 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 20550 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1628, + "step": 20551 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 20552 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 20553 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 20554 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 20555 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1527, + "step": 20556 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 20557 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 20558 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1677, + "step": 20559 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1614, + "step": 20560 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 20561 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 20562 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1674, + "step": 20563 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 20564 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1712, + "step": 20565 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1447, + "step": 20566 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 20567 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 20568 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.154, + "step": 20569 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 20570 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 20571 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 20572 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 20573 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 20574 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 20575 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 20576 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 20577 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 20578 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 20579 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1679, + "step": 20580 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 20581 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 20582 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 20583 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 20584 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1623, + "step": 20585 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 20586 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 20587 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 20588 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 20589 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 20590 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 20591 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 20592 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 20593 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 20594 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 20595 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 20596 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1728, + "step": 20597 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1606, + "step": 20598 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 20599 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 20600 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 20601 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 20602 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 20603 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1618, + "step": 20604 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 20605 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1776, + "step": 20606 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 20607 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1632, + "step": 20608 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 20609 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.0863, + "step": 20610 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 20611 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 20612 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 20613 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 20614 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1882, + "step": 20615 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 20616 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 20617 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 20618 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 20619 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1979, + "step": 20620 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 20621 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 20622 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 20623 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 20624 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.145, + "step": 20625 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1585, + "step": 20626 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1573, + "step": 20627 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1745, + "step": 20628 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 20629 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 20630 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 20631 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 20632 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 20633 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 20634 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1599, + "step": 20635 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1729, + "step": 20636 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 20637 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 20638 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 20639 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 20640 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 20641 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 20642 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 20643 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 20644 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 20645 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 20646 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1513, + "step": 20647 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1476, + "step": 20648 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 20649 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 20650 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1442, + "step": 20651 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 20652 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 20653 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 20654 + }, + { + "epoch": 1.25, + "learning_rate": 2e-05, + "loss": 0.1662, + "step": 20655 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 20656 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 20657 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 20658 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 20659 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 20660 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 20661 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 20662 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 20663 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 20664 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 20665 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 20666 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1476, + "step": 20667 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 20668 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1462, + "step": 20669 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 20670 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 20671 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1578, + "step": 20672 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1567, + "step": 20673 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 20674 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 20675 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 20676 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 20677 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 20678 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 20679 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 20680 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 20681 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 20682 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 20683 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 20684 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 20685 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 20686 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1442, + "step": 20687 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1518, + "step": 20688 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 20689 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1502, + "step": 20690 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1694, + "step": 20691 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 20692 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 20693 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 20694 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1568, + "step": 20695 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 20696 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 20697 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.158, + "step": 20698 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 20699 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 20700 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 20701 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 20702 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 20703 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 20704 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1967, + "step": 20705 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 20706 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 20707 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 20708 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 20709 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 20710 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 20711 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 20712 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 20713 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 20714 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 20715 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 20716 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 20717 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 20718 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 20719 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 20720 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 20721 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1629, + "step": 20722 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 20723 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 20724 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 20725 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 20726 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 20727 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 20728 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 20729 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 20730 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 20731 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 20732 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 20733 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 20734 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 20735 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 20736 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1518, + "step": 20737 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 20738 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 20739 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 20740 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1602, + "step": 20741 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 20742 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 20743 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1765, + "step": 20744 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 20745 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 20746 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1792, + "step": 20747 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 20748 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 20749 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 20750 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 20751 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 20752 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 20753 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 20754 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 20755 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 20756 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 20757 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 20758 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 20759 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 20760 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 20761 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 20762 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1596, + "step": 20763 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 20764 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 20765 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 20766 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 20767 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 20768 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 20769 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1596, + "step": 20770 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1612, + "step": 20771 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 20772 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 20773 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 20774 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1511, + "step": 20775 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 20776 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 20777 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 20778 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 20779 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 20780 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1497, + "step": 20781 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 20782 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 20783 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 20784 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 20785 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 20786 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 20787 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 20788 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 20789 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 20790 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 20791 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 20792 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1573, + "step": 20793 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1707, + "step": 20794 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 20795 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 20796 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1664, + "step": 20797 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1612, + "step": 20798 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 20799 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1794, + "step": 20800 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 20801 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 20802 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 20803 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 20804 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 20805 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 20806 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1495, + "step": 20807 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 20808 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 20809 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 20810 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1787, + "step": 20811 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 20812 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 20813 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1528, + "step": 20814 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1622, + "step": 20815 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 20816 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 20817 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 20818 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 20819 + }, + { + "epoch": 1.26, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 20820 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.0888, + "step": 20821 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 20822 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 20823 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 20824 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1442, + "step": 20825 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 20826 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 20827 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 20828 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 20829 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 20830 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 20831 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 20832 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 20833 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 20834 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 20835 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 20836 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 20837 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 20838 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1965, + "step": 20839 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 20840 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 20841 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 20842 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1735, + "step": 20843 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1513, + "step": 20844 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.163, + "step": 20845 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1894, + "step": 20846 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 20847 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1453, + "step": 20848 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 20849 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 20850 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 20851 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 20852 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 20853 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 20854 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 20855 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 20856 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 20857 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1794, + "step": 20858 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1634, + "step": 20859 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 20860 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 20861 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 20862 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 20863 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 20864 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 20865 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 20866 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 20867 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 20868 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 20869 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1767, + "step": 20870 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 20871 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 20872 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 20873 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 20874 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 20875 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 20876 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 20877 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1552, + "step": 20878 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1672, + "step": 20879 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 20880 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 20881 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1558, + "step": 20882 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.147, + "step": 20883 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 20884 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 20885 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 20886 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 20887 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1681, + "step": 20888 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 20889 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 20890 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 20891 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 20892 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 20893 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 20894 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 20895 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 20896 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 20897 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 20898 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 20899 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1716, + "step": 20900 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 20901 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1707, + "step": 20902 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1528, + "step": 20903 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 20904 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 20905 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 20906 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 20907 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 20908 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 20909 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 20910 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 20911 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1851, + "step": 20912 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1528, + "step": 20913 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 20914 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 20915 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 20916 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 20917 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1564, + "step": 20918 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 20919 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 20920 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1752, + "step": 20921 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.0882, + "step": 20922 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 20923 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 20924 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1878, + "step": 20925 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 20926 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 20927 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 20928 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 20929 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 20930 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1611, + "step": 20931 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 20932 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 20933 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1541, + "step": 20934 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.172, + "step": 20935 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 20936 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.158, + "step": 20937 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 20938 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 20939 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 20940 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 20941 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 20942 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1635, + "step": 20943 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 20944 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 20945 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 20946 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 20947 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 20948 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 20949 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 20950 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1517, + "step": 20951 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1663, + "step": 20952 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 20953 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1507, + "step": 20954 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 20955 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1814, + "step": 20956 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 20957 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 20958 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 20959 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1527, + "step": 20960 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 20961 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 20962 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 20963 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 20964 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1668, + "step": 20965 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 20966 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 20967 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.164, + "step": 20968 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 20969 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 20970 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 20971 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 20972 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1701, + "step": 20973 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 20974 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 20975 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 20976 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 20977 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 20978 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 20979 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 20980 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 20981 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 20982 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 20983 + }, + { + "epoch": 1.27, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 20984 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 20985 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.0878, + "step": 20986 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 20987 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1772, + "step": 20988 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1589, + "step": 20989 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 20990 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 20991 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 20992 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 20993 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1641, + "step": 20994 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1635, + "step": 20995 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 20996 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.145, + "step": 20997 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 20998 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1669, + "step": 20999 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 21000 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.16, + "step": 21001 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1638, + "step": 21002 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.0863, + "step": 21003 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1624, + "step": 21004 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 21005 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1621, + "step": 21006 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 21007 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 21008 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 21009 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 21010 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 21011 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 21012 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 21013 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1951, + "step": 21014 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 21015 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 21016 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 21017 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 21018 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 21019 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 21020 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 21021 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 21022 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.0902, + "step": 21023 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 21024 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 21025 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 21026 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.148, + "step": 21027 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 21028 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 21029 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 21030 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 21031 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 21032 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 21033 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 21034 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 21035 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 21036 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 21037 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 21038 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 21039 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 21040 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.188, + "step": 21041 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 21042 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 21043 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 21044 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 21045 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 21046 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 21047 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 21048 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 21049 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 21050 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1485, + "step": 21051 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1579, + "step": 21052 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 21053 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 21054 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 21055 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.147, + "step": 21056 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.175, + "step": 21057 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 21058 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 21059 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 21060 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 21061 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1707, + "step": 21062 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 21063 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 21064 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.148, + "step": 21065 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1691, + "step": 21066 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1507, + "step": 21067 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1429, + "step": 21068 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1656, + "step": 21069 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1521, + "step": 21070 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 21071 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 21072 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 21073 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 21074 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 21075 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 21076 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 21077 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 21078 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 21079 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 21080 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 21081 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 21082 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 21083 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 21084 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 21085 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 21086 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 21087 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 21088 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 21089 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 21090 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 21091 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1627, + "step": 21092 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1577, + "step": 21093 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 21094 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 21095 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 21096 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 21097 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1751, + "step": 21098 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1555, + "step": 21099 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 21100 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 21101 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.152, + "step": 21102 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 21103 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1521, + "step": 21104 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1621, + "step": 21105 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 21106 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 21107 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1714, + "step": 21108 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 21109 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 21110 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1918, + "step": 21111 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 21112 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 21113 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 21114 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1443, + "step": 21115 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1645, + "step": 21116 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 21117 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 21118 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 21119 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 21120 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1469, + "step": 21121 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 21122 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 21123 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 21124 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 21125 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1599, + "step": 21126 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 21127 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1525, + "step": 21128 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 21129 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 21130 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 21131 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 21132 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 21133 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 21134 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 21135 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 21136 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 21137 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 21138 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 21139 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 21140 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 21141 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1583, + "step": 21142 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 21143 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 21144 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 21145 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1598, + "step": 21146 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1866, + "step": 21147 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 21148 + }, + { + "epoch": 1.28, + "learning_rate": 2e-05, + "loss": 0.1918, + "step": 21149 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 21150 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1478, + "step": 21151 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1622, + "step": 21152 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1697, + "step": 21153 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 21154 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 21155 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1681, + "step": 21156 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 21157 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 21158 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 21159 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.2041, + "step": 21160 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 21161 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 21162 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1673, + "step": 21163 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 21164 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 21165 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 21166 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 21167 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 21168 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 21169 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 21170 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 21171 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 21172 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 21173 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1582, + "step": 21174 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 21175 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 21176 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 21177 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 21178 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 21179 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 21180 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1627, + "step": 21181 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1542, + "step": 21182 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1601, + "step": 21183 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 21184 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 21185 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 21186 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 21187 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 21188 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 21189 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1511, + "step": 21190 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 21191 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 21192 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 21193 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 21194 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 21195 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 21196 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 21197 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1527, + "step": 21198 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 21199 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 21200 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 21201 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 21202 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 21203 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 21204 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 21205 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 21206 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 21207 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 21208 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 21209 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 21210 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 21211 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 21212 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 21213 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 21214 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1803, + "step": 21215 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1773, + "step": 21216 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 21217 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1838, + "step": 21218 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1548, + "step": 21219 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 21220 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 21221 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 21222 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 21223 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1589, + "step": 21224 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.154, + "step": 21225 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 21226 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1776, + "step": 21227 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 21228 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1713, + "step": 21229 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1908, + "step": 21230 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 21231 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 21232 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 21233 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1765, + "step": 21234 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1505, + "step": 21235 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 21236 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 21237 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1469, + "step": 21238 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 21239 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 21240 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 21241 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1742, + "step": 21242 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 21243 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 21244 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 21245 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1688, + "step": 21246 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 21247 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 21248 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.165, + "step": 21249 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 21250 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 21251 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 21252 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 21253 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1555, + "step": 21254 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 21255 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 21256 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1662, + "step": 21257 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 21258 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 21259 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 21260 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1465, + "step": 21261 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 21262 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 21263 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1535, + "step": 21264 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1756, + "step": 21265 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 21266 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 21267 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1518, + "step": 21268 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 21269 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 21270 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 21271 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 21272 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 21273 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 21274 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 21275 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1949, + "step": 21276 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 21277 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 21278 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 21279 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 21280 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1608, + "step": 21281 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 21282 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 21283 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 21284 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 21285 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 21286 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 21287 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 21288 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 21289 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 21290 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1554, + "step": 21291 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 21292 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 21293 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 21294 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 21295 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 21296 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 21297 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 21298 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 21299 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 21300 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 21301 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 21302 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 21303 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 21304 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 21305 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 21306 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 21307 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 21308 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 21309 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 21310 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.0824, + "step": 21311 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 21312 + }, + { + "epoch": 1.29, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 21313 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 21314 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 21315 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 21316 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 21317 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 21318 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1498, + "step": 21319 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 21320 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 21321 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.144, + "step": 21322 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 21323 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 21324 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 21325 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 21326 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 21327 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 21328 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 21329 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 21330 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1631, + "step": 21331 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1526, + "step": 21332 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 21333 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 21334 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 21335 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1917, + "step": 21336 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 21337 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 21338 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 21339 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 21340 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1723, + "step": 21341 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 21342 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 21343 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 21344 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 21345 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1515, + "step": 21346 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 21347 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 21348 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 21349 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 21350 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 21351 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 21352 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 21353 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1837, + "step": 21354 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 21355 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 21356 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1708, + "step": 21357 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1828, + "step": 21358 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 21359 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 21360 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 21361 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 21362 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1451, + "step": 21363 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1604, + "step": 21364 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 21365 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 21366 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 21367 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.2178, + "step": 21368 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1621, + "step": 21369 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 21370 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1833, + "step": 21371 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 21372 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1682, + "step": 21373 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1681, + "step": 21374 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 21375 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 21376 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 21377 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1429, + "step": 21378 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 21379 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 21380 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 21381 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 21382 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1861, + "step": 21383 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 21384 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1875, + "step": 21385 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 21386 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1662, + "step": 21387 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 21388 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1625, + "step": 21389 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 21390 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 21391 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 21392 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 21393 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1507, + "step": 21394 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 21395 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 21396 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 21397 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 21398 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1563, + "step": 21399 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1524, + "step": 21400 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1451, + "step": 21401 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 21402 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 21403 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 21404 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 21405 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.158, + "step": 21406 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 21407 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 21408 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 21409 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 21410 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 21411 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.158, + "step": 21412 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 21413 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.2022, + "step": 21414 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 21415 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1618, + "step": 21416 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 21417 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 21418 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 21419 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 21420 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1596, + "step": 21421 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 21422 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1485, + "step": 21423 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 21424 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 21425 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 21426 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 21427 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 21428 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1468, + "step": 21429 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 21430 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 21431 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.162, + "step": 21432 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 21433 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 21434 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 21435 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 21436 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 21437 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1558, + "step": 21438 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1765, + "step": 21439 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 21440 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 21441 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 21442 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 21443 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1808, + "step": 21444 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 21445 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 21446 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 21447 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 21448 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1465, + "step": 21449 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 21450 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 21451 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1507, + "step": 21452 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 21453 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 21454 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1518, + "step": 21455 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 21456 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 21457 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 21458 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 21459 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 21460 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.144, + "step": 21461 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1478, + "step": 21462 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1503, + "step": 21463 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 21464 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 21465 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 21466 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 21467 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 21468 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 21469 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 21470 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 21471 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 21472 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1578, + "step": 21473 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 21474 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 21475 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1647, + "step": 21476 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 21477 + }, + { + "epoch": 1.3, + "learning_rate": 2e-05, + "loss": 0.1517, + "step": 21478 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.0684, + "step": 21479 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 21480 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 21481 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 21482 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 21483 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 21484 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 21485 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 21486 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 21487 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1667, + "step": 21488 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 21489 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 21490 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1758, + "step": 21491 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 21492 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 21493 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 21494 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 21495 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 21496 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1805, + "step": 21497 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 21498 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 21499 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1616, + "step": 21500 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 21501 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1476, + "step": 21502 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1576, + "step": 21503 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1767, + "step": 21504 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 21505 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 21506 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1754, + "step": 21507 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 21508 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 21509 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 21510 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.2018, + "step": 21511 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1517, + "step": 21512 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 21513 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1596, + "step": 21514 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 21515 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1616, + "step": 21516 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 21517 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1541, + "step": 21518 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 21519 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1778, + "step": 21520 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 21521 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 21522 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 21523 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 21524 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 21525 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1459, + "step": 21526 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 21527 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 21528 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1485, + "step": 21529 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 21530 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 21531 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 21532 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 21533 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1456, + "step": 21534 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 21535 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 21536 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1656, + "step": 21537 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 21538 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 21539 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 21540 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 21541 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 21542 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 21543 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1742, + "step": 21544 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1582, + "step": 21545 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1754, + "step": 21546 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 21547 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 21548 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 21549 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1548, + "step": 21550 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 21551 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 21552 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 21553 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 21554 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1599, + "step": 21555 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 21556 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 21557 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 21558 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1585, + "step": 21559 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 21560 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 21561 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 21562 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 21563 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 21564 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 21565 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 21566 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 21567 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 21568 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 21569 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 21570 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 21571 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 21572 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 21573 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 21574 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 21575 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.171, + "step": 21576 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1704, + "step": 21577 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 21578 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 21579 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 21580 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 21581 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 21582 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1716, + "step": 21583 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 21584 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 21585 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 21586 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 21587 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 21588 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1476, + "step": 21589 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 21590 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1865, + "step": 21591 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1666, + "step": 21592 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 21593 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.2139, + "step": 21594 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 21595 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1608, + "step": 21596 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1887, + "step": 21597 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1553, + "step": 21598 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 21599 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 21600 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 21601 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1641, + "step": 21602 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 21603 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 21604 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 21605 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 21606 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 21607 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 21608 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 21609 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 21610 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1618, + "step": 21611 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1946, + "step": 21612 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 21613 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 21614 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1442, + "step": 21615 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 21616 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 21617 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 21618 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 21619 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.16, + "step": 21620 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 21621 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 21622 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 21623 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 21624 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 21625 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 21626 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 21627 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1596, + "step": 21628 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1728, + "step": 21629 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 21630 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 21631 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 21632 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1577, + "step": 21633 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 21634 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 21635 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 21636 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1578, + "step": 21637 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 21638 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 21639 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 21640 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1729, + "step": 21641 + }, + { + "epoch": 1.31, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 21642 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 21643 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.179, + "step": 21644 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 21645 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 21646 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 21647 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 21648 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 21649 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 21650 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 21651 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 21652 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 21653 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 21654 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 21655 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 21656 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 21657 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1498, + "step": 21658 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 21659 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1605, + "step": 21660 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 21661 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.152, + "step": 21662 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 21663 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 21664 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 21665 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 21666 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 21667 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 21668 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 21669 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 21670 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 21671 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 21672 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 21673 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 21674 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 21675 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1914, + "step": 21676 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 21677 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1807, + "step": 21678 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 21679 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1668, + "step": 21680 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1749, + "step": 21681 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1821, + "step": 21682 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 21683 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 21684 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1563, + "step": 21685 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 21686 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 21687 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1652, + "step": 21688 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 21689 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 21690 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 21691 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 21692 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1845, + "step": 21693 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 21694 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 21695 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 21696 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 21697 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 21698 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 21699 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 21700 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 21701 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1507, + "step": 21702 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1695, + "step": 21703 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 21704 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1608, + "step": 21705 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 21706 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1597, + "step": 21707 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1652, + "step": 21708 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 21709 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1552, + "step": 21710 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 21711 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1664, + "step": 21712 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 21713 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 21714 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 21715 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 21716 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1718, + "step": 21717 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 21718 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 21719 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 21720 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 21721 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 21722 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 21723 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 21724 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 21725 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.0838, + "step": 21726 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 21727 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 21728 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 21729 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 21730 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 21731 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 21732 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1629, + "step": 21733 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 21734 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 21735 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 21736 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 21737 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 21738 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 21739 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 21740 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1971, + "step": 21741 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 21742 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 21743 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 21744 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 21745 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1524, + "step": 21746 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 21747 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 21748 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 21749 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1525, + "step": 21750 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1526, + "step": 21751 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1916, + "step": 21752 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 21753 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1907, + "step": 21754 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 21755 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 21756 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1516, + "step": 21757 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 21758 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.152, + "step": 21759 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 21760 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 21761 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 21762 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 21763 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 21764 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1669, + "step": 21765 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 21766 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 21767 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 21768 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1691, + "step": 21769 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1518, + "step": 21770 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1611, + "step": 21771 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1708, + "step": 21772 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1515, + "step": 21773 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 21774 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1462, + "step": 21775 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.0967, + "step": 21776 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1802, + "step": 21777 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 21778 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 21779 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 21780 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 21781 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 21782 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.191, + "step": 21783 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 21784 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 21785 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1497, + "step": 21786 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 21787 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 21788 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 21789 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1924, + "step": 21790 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 21791 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 21792 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 21793 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 21794 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 21795 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 21796 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1535, + "step": 21797 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 21798 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 21799 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 21800 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 21801 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 21802 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 21803 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1505, + "step": 21804 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 21805 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 21806 + }, + { + "epoch": 1.32, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 21807 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 21808 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 21809 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 21810 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1679, + "step": 21811 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 21812 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 21813 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 21814 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1656, + "step": 21815 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1723, + "step": 21816 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 21817 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1708, + "step": 21818 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 21819 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1485, + "step": 21820 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1657, + "step": 21821 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 21822 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1637, + "step": 21823 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.184, + "step": 21824 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 21825 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1631, + "step": 21826 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1694, + "step": 21827 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 21828 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1533, + "step": 21829 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 21830 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 21831 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1515, + "step": 21832 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 21833 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1677, + "step": 21834 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 21835 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1689, + "step": 21836 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1629, + "step": 21837 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 21838 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 21839 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1735, + "step": 21840 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 21841 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 21842 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 21843 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 21844 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1716, + "step": 21845 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 21846 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 21847 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 21848 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 21849 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 21850 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 21851 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 21852 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 21853 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 21854 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 21855 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1497, + "step": 21856 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1747, + "step": 21857 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1801, + "step": 21858 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1578, + "step": 21859 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 21860 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 21861 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1516, + "step": 21862 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 21863 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1505, + "step": 21864 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 21865 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1669, + "step": 21866 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 21867 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 21868 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.156, + "step": 21869 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 21870 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 21871 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 21872 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1447, + "step": 21873 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 21874 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1516, + "step": 21875 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 21876 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.183, + "step": 21877 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 21878 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 21879 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 21880 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 21881 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 21882 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 21883 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 21884 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 21885 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1622, + "step": 21886 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 21887 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1595, + "step": 21888 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 21889 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 21890 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 21891 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 21892 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1538, + "step": 21893 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 21894 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1632, + "step": 21895 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1679, + "step": 21896 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 21897 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 21898 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 21899 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1659, + "step": 21900 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 21901 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 21902 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1561, + "step": 21903 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1919, + "step": 21904 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 21905 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 21906 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 21907 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1631, + "step": 21908 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 21909 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1429, + "step": 21910 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 21911 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 21912 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1531, + "step": 21913 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 21914 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 21915 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 21916 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 21917 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1518, + "step": 21918 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1605, + "step": 21919 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1669, + "step": 21920 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 21921 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 21922 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 21923 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 21924 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 21925 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1745, + "step": 21926 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 21927 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 21928 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 21929 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 21930 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 21931 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 21932 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 21933 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1659, + "step": 21934 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 21935 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 21936 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 21937 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 21938 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 21939 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 21940 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 21941 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 21942 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 21943 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1429, + "step": 21944 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 21945 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 21946 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 21947 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 21948 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 21949 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 21950 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 21951 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1568, + "step": 21952 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 21953 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 21954 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 21955 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 21956 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 21957 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1677, + "step": 21958 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 21959 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1502, + "step": 21960 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1621, + "step": 21961 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1647, + "step": 21962 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 21963 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 21964 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1517, + "step": 21965 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.0936, + "step": 21966 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 21967 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1546, + "step": 21968 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 21969 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 21970 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 21971 + }, + { + "epoch": 1.33, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 21972 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 21973 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 21974 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 21975 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 21976 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 21977 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 21978 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 21979 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 21980 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 21981 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 21982 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 21983 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 21984 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 21985 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 21986 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 21987 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1763, + "step": 21988 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 21989 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 21990 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 21991 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 21992 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1582, + "step": 21993 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 21994 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 21995 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 21996 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1681, + "step": 21997 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1669, + "step": 21998 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1657, + "step": 21999 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1782, + "step": 22000 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 22001 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 22002 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 22003 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 22004 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 22005 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 22006 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 22007 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 22008 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1628, + "step": 22009 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 22010 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 22011 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 22012 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1731, + "step": 22013 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.183, + "step": 22014 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 22015 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 22016 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 22017 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1798, + "step": 22018 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 22019 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 22020 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 22021 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 22022 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1462, + "step": 22023 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 22024 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 22025 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 22026 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 22027 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 22028 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 22029 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1679, + "step": 22030 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1643, + "step": 22031 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 22032 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 22033 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 22034 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 22035 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 22036 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 22037 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 22038 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 22039 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 22040 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 22041 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1667, + "step": 22042 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 22043 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 22044 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 22045 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 22046 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 22047 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1702, + "step": 22048 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1563, + "step": 22049 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 22050 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 22051 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 22052 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 22053 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 22054 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 22055 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 22056 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 22057 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 22058 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 22059 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 22060 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 22061 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 22062 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 22063 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 22064 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 22065 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 22066 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 22067 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 22068 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 22069 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 22070 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 22071 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 22072 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 22073 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1497, + "step": 22074 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 22075 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 22076 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1528, + "step": 22077 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1502, + "step": 22078 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 22079 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 22080 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 22081 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1498, + "step": 22082 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 22083 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 22084 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1867, + "step": 22085 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 22086 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 22087 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.0935, + "step": 22088 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1895, + "step": 22089 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 22090 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 22091 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1642, + "step": 22092 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 22093 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 22094 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 22095 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 22096 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 22097 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 22098 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 22099 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 22100 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1608, + "step": 22101 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 22102 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 22103 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 22104 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 22105 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1616, + "step": 22106 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 22107 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 22108 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 22109 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.165, + "step": 22110 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 22111 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1654, + "step": 22112 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 22113 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.147, + "step": 22114 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 22115 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 22116 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 22117 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 22118 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 22119 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 22120 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 22121 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1805, + "step": 22122 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 22123 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 22124 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 22125 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 22126 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 22127 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 22128 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 22129 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 22130 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 22131 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 22132 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 22133 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 22134 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 22135 + }, + { + "epoch": 1.34, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 22136 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 22137 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 22138 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 22139 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1628, + "step": 22140 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1786, + "step": 22141 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1643, + "step": 22142 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 22143 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 22144 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 22145 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 22146 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 22147 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 22148 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 22149 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 22150 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1516, + "step": 22151 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 22152 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 22153 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 22154 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 22155 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1598, + "step": 22156 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 22157 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 22158 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 22159 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 22160 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 22161 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 22162 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 22163 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 22164 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 22165 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1607, + "step": 22166 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 22167 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 22168 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 22169 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 22170 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 22171 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 22172 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1932, + "step": 22173 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 22174 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 22175 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 22176 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 22177 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 22178 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 22179 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 22180 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 22181 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 22182 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 22183 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 22184 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 22185 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 22186 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 22187 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 22188 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 22189 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 22190 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1535, + "step": 22191 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1604, + "step": 22192 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 22193 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 22194 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 22195 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 22196 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 22197 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 22198 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 22199 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 22200 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 22201 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 22202 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 22203 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1657, + "step": 22204 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 22205 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 22206 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 22207 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 22208 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1705, + "step": 22209 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 22210 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 22211 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 22212 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 22213 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 22214 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 22215 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 22216 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 22217 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 22218 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 22219 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 22220 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1599, + "step": 22221 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 22222 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 22223 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 22224 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 22225 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1611, + "step": 22226 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 22227 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 22228 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 22229 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 22230 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 22231 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 22232 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1716, + "step": 22233 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 22234 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 22235 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 22236 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 22237 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 22238 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1665, + "step": 22239 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1845, + "step": 22240 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1628, + "step": 22241 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1774, + "step": 22242 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1621, + "step": 22243 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 22244 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 22245 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1851, + "step": 22246 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1505, + "step": 22247 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 22248 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1625, + "step": 22249 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 22250 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1442, + "step": 22251 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 22252 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 22253 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 22254 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 22255 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1505, + "step": 22256 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1517, + "step": 22257 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 22258 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 22259 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 22260 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 22261 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1647, + "step": 22262 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1552, + "step": 22263 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 22264 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.152, + "step": 22265 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 22266 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 22267 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1846, + "step": 22268 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 22269 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1791, + "step": 22270 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 22271 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 22272 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.0899, + "step": 22273 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 22274 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 22275 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 22276 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 22277 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 22278 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 22279 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 22280 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1552, + "step": 22281 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.2092, + "step": 22282 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 22283 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 22284 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 22285 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 22286 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 22287 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1686, + "step": 22288 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 22289 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 22290 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 22291 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 22292 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1612, + "step": 22293 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 22294 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1884, + "step": 22295 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 22296 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 22297 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 22298 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 22299 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 22300 + }, + { + "epoch": 1.35, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 22301 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 22302 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 22303 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 22304 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 22305 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 22306 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 22307 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 22308 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 22309 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 22310 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 22311 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 22312 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 22313 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 22314 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 22315 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 22316 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 22317 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 22318 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.155, + "step": 22319 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 22320 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 22321 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 22322 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 22323 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 22324 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 22325 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 22326 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 22327 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 22328 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1738, + "step": 22329 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 22330 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 22331 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 22332 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 22333 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1701, + "step": 22334 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 22335 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 22336 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 22337 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 22338 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 22339 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 22340 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 22341 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 22342 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 22343 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 22344 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 22345 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 22346 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 22347 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 22348 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 22349 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 22350 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 22351 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1578, + "step": 22352 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 22353 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1856, + "step": 22354 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 22355 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 22356 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 22357 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 22358 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1804, + "step": 22359 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 22360 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.156, + "step": 22361 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 22362 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1597, + "step": 22363 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 22364 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 22365 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.2157, + "step": 22366 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 22367 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 22368 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 22369 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 22370 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.193, + "step": 22371 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1563, + "step": 22372 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 22373 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 22374 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 22375 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1659, + "step": 22376 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1959, + "step": 22377 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1612, + "step": 22378 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 22379 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 22380 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 22381 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 22382 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1617, + "step": 22383 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1919, + "step": 22384 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 22385 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 22386 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.145, + "step": 22387 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 22388 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 22389 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 22390 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 22391 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 22392 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 22393 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1476, + "step": 22394 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 22395 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 22396 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1647, + "step": 22397 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 22398 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 22399 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 22400 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 22401 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 22402 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1854, + "step": 22403 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 22404 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 22405 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 22406 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 22407 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 22408 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1673, + "step": 22409 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 22410 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 22411 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1835, + "step": 22412 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 22413 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 22414 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 22415 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1549, + "step": 22416 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 22417 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 22418 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1535, + "step": 22419 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 22420 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 22421 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 22422 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 22423 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 22424 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.205, + "step": 22425 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 22426 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 22427 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1826, + "step": 22428 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 22429 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 22430 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 22431 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 22432 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1791, + "step": 22433 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 22434 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 22435 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 22436 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 22437 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 22438 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 22439 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1761, + "step": 22440 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 22441 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 22442 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 22443 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 22444 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 22445 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 22446 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 22447 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1499, + "step": 22448 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 22449 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 22450 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 22451 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 22452 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 22453 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 22454 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 22455 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1704, + "step": 22456 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1624, + "step": 22457 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 22458 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 22459 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1707, + "step": 22460 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1495, + "step": 22461 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 22462 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1784, + "step": 22463 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 22464 + }, + { + "epoch": 1.36, + "learning_rate": 2e-05, + "loss": 0.1732, + "step": 22465 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 22466 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 22467 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.144, + "step": 22468 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1701, + "step": 22469 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 22470 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1507, + "step": 22471 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 22472 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 22473 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 22474 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1554, + "step": 22475 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1709, + "step": 22476 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 22477 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 22478 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 22479 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 22480 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 22481 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 22482 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 22483 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 22484 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 22485 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 22486 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.2065, + "step": 22487 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 22488 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 22489 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 22490 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 22491 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 22492 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 22493 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 22494 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 22495 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1586, + "step": 22496 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 22497 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 22498 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1675, + "step": 22499 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1757, + "step": 22500 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 22501 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 22502 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 22503 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 22504 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 22505 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1598, + "step": 22506 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 22507 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 22508 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 22509 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 22510 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.145, + "step": 22511 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 22512 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1531, + "step": 22513 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1663, + "step": 22514 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 22515 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 22516 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 22517 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 22518 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 22519 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.178, + "step": 22520 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 22521 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1726, + "step": 22522 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 22523 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 22524 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 22525 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1527, + "step": 22526 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 22527 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 22528 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 22529 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 22530 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 22531 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1519, + "step": 22532 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 22533 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 22534 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1656, + "step": 22535 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 22536 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 22537 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 22538 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1485, + "step": 22539 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 22540 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 22541 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 22542 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 22543 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 22544 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1476, + "step": 22545 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.0952, + "step": 22546 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 22547 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1525, + "step": 22548 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 22549 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 22550 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 22551 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 22552 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1525, + "step": 22553 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 22554 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 22555 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1897, + "step": 22556 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 22557 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 22558 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 22559 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 22560 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 22561 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1649, + "step": 22562 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 22563 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 22564 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 22565 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 22566 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 22567 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1513, + "step": 22568 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 22569 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 22570 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1674, + "step": 22571 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1629, + "step": 22572 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 22573 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1831, + "step": 22574 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 22575 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 22576 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 22577 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 22578 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.154, + "step": 22579 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 22580 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.164, + "step": 22581 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 22582 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 22583 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 22584 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 22585 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 22586 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1534, + "step": 22587 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 22588 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 22589 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 22590 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 22591 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 22592 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 22593 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 22594 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 22595 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 22596 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1574, + "step": 22597 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 22598 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 22599 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 22600 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 22601 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 22602 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 22603 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 22604 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 22605 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1919, + "step": 22606 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 22607 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 22608 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 22609 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 22610 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 22611 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1695, + "step": 22612 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 22613 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.167, + "step": 22614 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1665, + "step": 22615 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 22616 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 22617 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 22618 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 22619 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1605, + "step": 22620 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 22621 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 22622 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 22623 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 22624 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 22625 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 22626 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1627, + "step": 22627 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1671, + "step": 22628 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.1521, + "step": 22629 + }, + { + "epoch": 1.37, + "learning_rate": 2e-05, + "loss": 0.145, + "step": 22630 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1621, + "step": 22631 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 22632 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 22633 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 22634 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 22635 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 22636 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 22637 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1827, + "step": 22638 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 22639 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 22640 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 22641 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 22642 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 22643 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 22644 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1824, + "step": 22645 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 22646 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 22647 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1488, + "step": 22648 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 22649 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 22650 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1713, + "step": 22651 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 22652 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 22653 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 22654 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 22655 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1574, + "step": 22656 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 22657 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 22658 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 22659 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 22660 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 22661 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1754, + "step": 22662 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1733, + "step": 22663 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 22664 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1442, + "step": 22665 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1533, + "step": 22666 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 22667 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 22668 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 22669 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 22670 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 22671 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1657, + "step": 22672 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1495, + "step": 22673 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 22674 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1552, + "step": 22675 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 22676 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 22677 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 22678 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 22679 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.144, + "step": 22680 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 22681 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 22682 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 22683 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1519, + "step": 22684 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 22685 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 22686 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 22687 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 22688 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 22689 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 22690 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 22691 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1708, + "step": 22692 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 22693 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 22694 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 22695 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 22696 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 22697 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1783, + "step": 22698 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1546, + "step": 22699 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 22700 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1502, + "step": 22701 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 22702 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 22703 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1617, + "step": 22704 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 22705 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 22706 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 22707 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1614, + "step": 22708 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 22709 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 22710 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 22711 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 22712 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1606, + "step": 22713 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 22714 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 22715 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 22716 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 22717 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 22718 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 22719 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 22720 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 22721 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 22722 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 22723 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1751, + "step": 22724 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1687, + "step": 22725 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 22726 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 22727 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 22728 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 22729 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.0914, + "step": 22730 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1724, + "step": 22731 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 22732 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 22733 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1629, + "step": 22734 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 22735 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1456, + "step": 22736 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 22737 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1606, + "step": 22738 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 22739 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1829, + "step": 22740 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1679, + "step": 22741 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 22742 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1553, + "step": 22743 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 22744 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.148, + "step": 22745 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 22746 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 22747 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1685, + "step": 22748 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 22749 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1497, + "step": 22750 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 22751 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 22752 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 22753 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 22754 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.163, + "step": 22755 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.144, + "step": 22756 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 22757 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.172, + "step": 22758 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 22759 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1429, + "step": 22760 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 22761 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 22762 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 22763 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.0855, + "step": 22764 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 22765 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 22766 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1669, + "step": 22767 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 22768 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 22769 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 22770 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 22771 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 22772 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 22773 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 22774 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 22775 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 22776 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 22777 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 22778 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1663, + "step": 22779 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 22780 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 22781 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.163, + "step": 22782 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 22783 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 22784 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 22785 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 22786 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 22787 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 22788 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 22789 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 22790 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 22791 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 22792 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 22793 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 22794 + }, + { + "epoch": 1.38, + "learning_rate": 2e-05, + "loss": 0.1928, + "step": 22795 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1485, + "step": 22796 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1687, + "step": 22797 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 22798 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 22799 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 22800 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 22801 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.16, + "step": 22802 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1505, + "step": 22803 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 22804 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.0891, + "step": 22805 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 22806 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 22807 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 22808 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 22809 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1762, + "step": 22810 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1724, + "step": 22811 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 22812 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1606, + "step": 22813 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1551, + "step": 22814 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 22815 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 22816 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1612, + "step": 22817 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 22818 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 22819 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 22820 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.0929, + "step": 22821 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 22822 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.167, + "step": 22823 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 22824 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 22825 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 22826 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 22827 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 22828 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 22829 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 22830 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1451, + "step": 22831 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 22832 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 22833 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 22834 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.153, + "step": 22835 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1507, + "step": 22836 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1443, + "step": 22837 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 22838 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1531, + "step": 22839 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 22840 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1564, + "step": 22841 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 22842 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1495, + "step": 22843 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 22844 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 22845 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 22846 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 22847 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1659, + "step": 22848 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 22849 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 22850 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1639, + "step": 22851 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1582, + "step": 22852 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1766, + "step": 22853 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 22854 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 22855 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 22856 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 22857 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.2016, + "step": 22858 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 22859 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1687, + "step": 22860 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.0837, + "step": 22861 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 22862 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 22863 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 22864 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 22865 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 22866 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 22867 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1628, + "step": 22868 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 22869 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 22870 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 22871 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 22872 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 22873 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 22874 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1557, + "step": 22875 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 22876 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 22877 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 22878 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.175, + "step": 22879 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1723, + "step": 22880 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 22881 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 22882 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 22883 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 22884 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1573, + "step": 22885 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 22886 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 22887 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.0766, + "step": 22888 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.147, + "step": 22889 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 22890 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 22891 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 22892 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 22893 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 22894 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 22895 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 22896 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 22897 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 22898 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 22899 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 22900 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.0819, + "step": 22901 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 22902 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 22903 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 22904 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 22905 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.161, + "step": 22906 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 22907 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 22908 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 22909 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 22910 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 22911 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 22912 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1747, + "step": 22913 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 22914 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 22915 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1548, + "step": 22916 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1585, + "step": 22917 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 22918 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 22919 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 22920 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 22921 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 22922 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 22923 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 22924 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 22925 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 22926 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 22927 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 22928 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 22929 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 22930 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 22931 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 22932 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 22933 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 22934 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1578, + "step": 22935 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1676, + "step": 22936 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1768, + "step": 22937 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 22938 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1926, + "step": 22939 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 22940 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 22941 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 22942 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 22943 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 22944 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.0943, + "step": 22945 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 22946 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1567, + "step": 22947 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 22948 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 22949 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1495, + "step": 22950 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 22951 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 22952 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 22953 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 22954 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 22955 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 22956 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1495, + "step": 22957 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.1681, + "step": 22958 + }, + { + "epoch": 1.39, + "learning_rate": 2e-05, + "loss": 0.173, + "step": 22959 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 22960 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.0952, + "step": 22961 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 22962 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.188, + "step": 22963 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 22964 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 22965 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 22966 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 22967 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 22968 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 22969 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 22970 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 22971 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 22972 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 22973 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 22974 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 22975 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 22976 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 22977 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 22978 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 22979 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 22980 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.169, + "step": 22981 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 22982 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1662, + "step": 22983 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.153, + "step": 22984 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1685, + "step": 22985 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 22986 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 22987 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 22988 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1726, + "step": 22989 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 22990 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 22991 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 22992 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1534, + "step": 22993 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 22994 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 22995 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1603, + "step": 22996 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 22997 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1551, + "step": 22998 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 22999 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.153, + "step": 23000 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 23001 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.148, + "step": 23002 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 23003 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 23004 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 23005 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 23006 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1625, + "step": 23007 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 23008 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 23009 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 23010 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 23011 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 23012 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 23013 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 23014 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 23015 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1631, + "step": 23016 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 23017 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 23018 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 23019 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 23020 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 23021 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1873, + "step": 23022 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 23023 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.2162, + "step": 23024 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1752, + "step": 23025 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1763, + "step": 23026 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.146, + "step": 23027 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1612, + "step": 23028 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1597, + "step": 23029 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 23030 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 23031 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 23032 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 23033 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 23034 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1531, + "step": 23035 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 23036 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 23037 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 23038 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1573, + "step": 23039 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 23040 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 23041 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 23042 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 23043 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 23044 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 23045 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 23046 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 23047 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 23048 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 23049 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 23050 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 23051 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 23052 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 23053 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1698, + "step": 23054 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1701, + "step": 23055 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 23056 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 23057 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 23058 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 23059 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 23060 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 23061 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 23062 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 23063 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 23064 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 23065 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 23066 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 23067 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 23068 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 23069 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1595, + "step": 23070 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1646, + "step": 23071 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1656, + "step": 23072 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 23073 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1618, + "step": 23074 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 23075 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 23076 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 23077 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1644, + "step": 23078 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 23079 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 23080 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 23081 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 23082 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 23083 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 23084 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1596, + "step": 23085 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 23086 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 23087 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 23088 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1505, + "step": 23089 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 23090 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 23091 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1823, + "step": 23092 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 23093 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 23094 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1658, + "step": 23095 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 23096 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 23097 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 23098 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 23099 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 23100 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 23101 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 23102 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 23103 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 23104 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.0902, + "step": 23105 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1758, + "step": 23106 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 23107 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 23108 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 23109 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 23110 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 23111 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 23112 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 23113 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.0801, + "step": 23114 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 23115 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 23116 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 23117 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 23118 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 23119 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 23120 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 23121 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 23122 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 23123 + }, + { + "epoch": 1.4, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 23124 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 23125 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 23126 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 23127 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 23128 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 23129 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 23130 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1503, + "step": 23131 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1498, + "step": 23132 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 23133 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1647, + "step": 23134 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1502, + "step": 23135 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.2031, + "step": 23136 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 23137 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 23138 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 23139 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1498, + "step": 23140 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 23141 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 23142 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 23143 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 23144 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 23145 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 23146 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 23147 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 23148 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 23149 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 23150 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.153, + "step": 23151 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 23152 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 23153 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.0843, + "step": 23154 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 23155 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 23156 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 23157 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 23158 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 23159 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 23160 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 23161 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 23162 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 23163 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1607, + "step": 23164 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 23165 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1607, + "step": 23166 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 23167 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1966, + "step": 23168 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1797, + "step": 23169 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1563, + "step": 23170 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1542, + "step": 23171 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 23172 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 23173 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1591, + "step": 23174 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1531, + "step": 23175 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1678, + "step": 23176 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1804, + "step": 23177 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 23178 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 23179 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1573, + "step": 23180 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1602, + "step": 23181 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 23182 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 23183 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1601, + "step": 23184 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.0886, + "step": 23185 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1616, + "step": 23186 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 23187 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 23188 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 23189 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1818, + "step": 23190 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 23191 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 23192 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 23193 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1928, + "step": 23194 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 23195 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 23196 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 23197 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.0818, + "step": 23198 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1597, + "step": 23199 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1699, + "step": 23200 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 23201 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 23202 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.159, + "step": 23203 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 23204 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 23205 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 23206 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 23207 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 23208 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 23209 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1751, + "step": 23210 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 23211 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 23212 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 23213 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 23214 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1671, + "step": 23215 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 23216 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 23217 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1641, + "step": 23218 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 23219 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 23220 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 23221 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 23222 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1601, + "step": 23223 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 23224 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 23225 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 23226 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 23227 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 23228 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 23229 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 23230 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1655, + "step": 23231 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 23232 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1857, + "step": 23233 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 23234 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 23235 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1469, + "step": 23236 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1655, + "step": 23237 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 23238 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 23239 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 23240 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 23241 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1528, + "step": 23242 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 23243 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 23244 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 23245 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 23246 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 23247 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 23248 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.176, + "step": 23249 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1528, + "step": 23250 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1705, + "step": 23251 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1534, + "step": 23252 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 23253 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1453, + "step": 23254 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 23255 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 23256 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.153, + "step": 23257 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 23258 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 23259 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 23260 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1683, + "step": 23261 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1443, + "step": 23262 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 23263 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 23264 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1459, + "step": 23265 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 23266 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 23267 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 23268 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1518, + "step": 23269 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 23270 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 23271 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 23272 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 23273 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 23274 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 23275 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 23276 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 23277 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1459, + "step": 23278 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 23279 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 23280 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 23281 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1646, + "step": 23282 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 23283 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1613, + "step": 23284 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1633, + "step": 23285 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1603, + "step": 23286 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 23287 + }, + { + "epoch": 1.41, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 23288 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 23289 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 23290 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 23291 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 23292 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 23293 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 23294 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 23295 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 23296 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 23297 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 23298 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1671, + "step": 23299 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 23300 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 23301 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 23302 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 23303 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 23304 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.169, + "step": 23305 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 23306 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 23307 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 23308 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 23309 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1443, + "step": 23310 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 23311 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 23312 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 23313 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 23314 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 23315 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 23316 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 23317 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1515, + "step": 23318 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 23319 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 23320 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1626, + "step": 23321 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1453, + "step": 23322 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 23323 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 23324 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 23325 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 23326 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1534, + "step": 23327 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 23328 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 23329 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 23330 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 23331 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 23332 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 23333 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1549, + "step": 23334 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1955, + "step": 23335 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1837, + "step": 23336 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.0911, + "step": 23337 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 23338 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 23339 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 23340 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 23341 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1542, + "step": 23342 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 23343 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.145, + "step": 23344 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1602, + "step": 23345 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 23346 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1585, + "step": 23347 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.144, + "step": 23348 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 23349 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1465, + "step": 23350 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 23351 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 23352 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 23353 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 23354 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1576, + "step": 23355 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 23356 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 23357 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 23358 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 23359 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 23360 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 23361 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 23362 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 23363 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 23364 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 23365 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.163, + "step": 23366 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 23367 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 23368 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 23369 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 23370 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 23371 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 23372 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 23373 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 23374 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 23375 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 23376 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 23377 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 23378 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 23379 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1807, + "step": 23380 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 23381 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 23382 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1944, + "step": 23383 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 23384 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 23385 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 23386 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 23387 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 23388 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1577, + "step": 23389 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 23390 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 23391 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 23392 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 23393 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 23394 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.162, + "step": 23395 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1548, + "step": 23396 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 23397 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 23398 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 23399 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 23400 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1546, + "step": 23401 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1749, + "step": 23402 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 23403 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 23404 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 23405 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 23406 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 23407 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1459, + "step": 23408 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 23409 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1819, + "step": 23410 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 23411 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 23412 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 23413 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 23414 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1476, + "step": 23415 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1758, + "step": 23416 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 23417 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 23418 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 23419 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 23420 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 23421 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 23422 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 23423 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 23424 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1516, + "step": 23425 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 23426 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 23427 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 23428 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.167, + "step": 23429 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.195, + "step": 23430 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 23431 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 23432 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.155, + "step": 23433 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 23434 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1733, + "step": 23435 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1915, + "step": 23436 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 23437 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1515, + "step": 23438 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 23439 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.146, + "step": 23440 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 23441 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 23442 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 23443 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1551, + "step": 23444 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 23445 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1453, + "step": 23446 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1664, + "step": 23447 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 23448 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 23449 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 23450 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1696, + "step": 23451 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 23452 + }, + { + "epoch": 1.42, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 23453 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 23454 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 23455 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 23456 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1443, + "step": 23457 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 23458 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1498, + "step": 23459 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1731, + "step": 23460 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 23461 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 23462 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1794, + "step": 23463 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.16, + "step": 23464 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 23465 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1655, + "step": 23466 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 23467 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 23468 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 23469 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 23470 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 23471 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 23472 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 23473 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 23474 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1638, + "step": 23475 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 23476 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 23477 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 23478 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1451, + "step": 23479 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 23480 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1516, + "step": 23481 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 23482 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1773, + "step": 23483 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1462, + "step": 23484 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 23485 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 23486 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 23487 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1872, + "step": 23488 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 23489 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 23490 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1462, + "step": 23491 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 23492 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 23493 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1485, + "step": 23494 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 23495 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 23496 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 23497 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.147, + "step": 23498 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1689, + "step": 23499 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 23500 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 23501 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 23502 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1579, + "step": 23503 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.181, + "step": 23504 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 23505 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1684, + "step": 23506 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1585, + "step": 23507 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 23508 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 23509 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 23510 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1668, + "step": 23511 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 23512 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 23513 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 23514 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 23515 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 23516 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 23517 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1588, + "step": 23518 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 23519 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 23520 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 23521 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 23522 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 23523 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1574, + "step": 23524 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 23525 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 23526 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 23527 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 23528 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 23529 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 23530 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 23531 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1745, + "step": 23532 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1503, + "step": 23533 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1485, + "step": 23534 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 23535 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 23536 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 23537 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 23538 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 23539 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 23540 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 23541 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 23542 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 23543 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 23544 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 23545 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 23546 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1582, + "step": 23547 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 23548 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1757, + "step": 23549 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.0929, + "step": 23550 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 23551 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 23552 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 23553 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 23554 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 23555 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 23556 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 23557 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1533, + "step": 23558 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 23559 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 23560 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.186, + "step": 23561 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 23562 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 23563 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1701, + "step": 23564 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1576, + "step": 23565 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 23566 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 23567 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 23568 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 23569 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 23570 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 23571 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 23572 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 23573 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.153, + "step": 23574 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 23575 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 23576 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 23577 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 23578 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 23579 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1771, + "step": 23580 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1488, + "step": 23581 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 23582 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 23583 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1645, + "step": 23584 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 23585 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 23586 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 23587 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 23588 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 23589 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 23590 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 23591 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 23592 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.159, + "step": 23593 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 23594 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 23595 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 23596 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 23597 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 23598 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 23599 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 23600 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 23601 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1488, + "step": 23602 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1596, + "step": 23603 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1558, + "step": 23604 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 23605 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1586, + "step": 23606 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 23607 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1736, + "step": 23608 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 23609 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.182, + "step": 23610 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 23611 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 23612 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 23613 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.174, + "step": 23614 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 23615 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.167, + "step": 23616 + }, + { + "epoch": 1.43, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 23617 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 23618 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 23619 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1546, + "step": 23620 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1567, + "step": 23621 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1686, + "step": 23622 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 23623 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1566, + "step": 23624 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 23625 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 23626 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 23627 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1728, + "step": 23628 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 23629 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1447, + "step": 23630 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.167, + "step": 23631 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 23632 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 23633 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 23634 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 23635 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 23636 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 23637 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 23638 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 23639 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1649, + "step": 23640 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1582, + "step": 23641 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 23642 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 23643 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 23644 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 23645 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 23646 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 23647 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 23648 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 23649 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 23650 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1591, + "step": 23651 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1532, + "step": 23652 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1779, + "step": 23653 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 23654 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 23655 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 23656 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 23657 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 23658 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 23659 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 23660 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 23661 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 23662 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 23663 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 23664 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.176, + "step": 23665 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 23666 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.188, + "step": 23667 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 23668 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 23669 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 23670 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 23671 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 23672 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 23673 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 23674 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 23675 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1698, + "step": 23676 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 23677 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 23678 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 23679 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.177, + "step": 23680 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1973, + "step": 23681 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1453, + "step": 23682 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 23683 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 23684 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 23685 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 23686 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 23687 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1688, + "step": 23688 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 23689 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 23690 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 23691 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 23692 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.164, + "step": 23693 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 23694 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1538, + "step": 23695 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 23696 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 23697 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.083, + "step": 23698 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 23699 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 23700 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1518, + "step": 23701 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 23702 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 23703 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1626, + "step": 23704 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1503, + "step": 23705 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 23706 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 23707 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 23708 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1707, + "step": 23709 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1599, + "step": 23710 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 23711 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1579, + "step": 23712 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 23713 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 23714 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 23715 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 23716 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1618, + "step": 23717 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 23718 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 23719 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.152, + "step": 23720 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 23721 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1858, + "step": 23722 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 23723 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 23724 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 23725 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1564, + "step": 23726 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 23727 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1564, + "step": 23728 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 23729 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 23730 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 23731 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 23732 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1786, + "step": 23733 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.152, + "step": 23734 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 23735 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 23736 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1468, + "step": 23737 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 23738 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 23739 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 23740 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 23741 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 23742 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 23743 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 23744 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 23745 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 23746 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 23747 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 23748 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 23749 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 23750 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1704, + "step": 23751 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 23752 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 23753 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 23754 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 23755 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 23756 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 23757 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.159, + "step": 23758 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 23759 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 23760 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1488, + "step": 23761 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1447, + "step": 23762 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 23763 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1642, + "step": 23764 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1707, + "step": 23765 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 23766 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 23767 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 23768 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 23769 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 23770 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1722, + "step": 23771 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 23772 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 23773 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1754, + "step": 23774 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 23775 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 23776 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 23777 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 23778 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 23779 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 23780 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 23781 + }, + { + "epoch": 1.44, + "learning_rate": 2e-05, + "loss": 0.087, + "step": 23782 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 23783 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 23784 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 23785 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 23786 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 23787 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 23788 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1591, + "step": 23789 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 23790 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 23791 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 23792 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1505, + "step": 23793 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1771, + "step": 23794 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 23795 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1533, + "step": 23796 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 23797 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 23798 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 23799 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1892, + "step": 23800 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1462, + "step": 23801 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 23802 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 23803 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1716, + "step": 23804 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1631, + "step": 23805 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 23806 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 23807 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 23808 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 23809 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 23810 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1566, + "step": 23811 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1816, + "step": 23812 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 23813 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 23814 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 23815 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1601, + "step": 23816 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1714, + "step": 23817 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 23818 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 23819 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 23820 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1579, + "step": 23821 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 23822 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1642, + "step": 23823 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 23824 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1538, + "step": 23825 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 23826 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 23827 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 23828 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 23829 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 23830 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 23831 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 23832 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 23833 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 23834 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 23835 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1785, + "step": 23836 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.0936, + "step": 23837 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 23838 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 23839 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 23840 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 23841 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 23842 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1793, + "step": 23843 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 23844 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 23845 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 23846 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 23847 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1541, + "step": 23848 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1787, + "step": 23849 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 23850 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 23851 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 23852 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 23853 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 23854 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 23855 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1644, + "step": 23856 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 23857 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 23858 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.0876, + "step": 23859 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 23860 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 23861 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 23862 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 23863 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 23864 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 23865 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 23866 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 23867 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 23868 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1527, + "step": 23869 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 23870 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 23871 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 23872 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 23873 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 23874 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 23875 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1429, + "step": 23876 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 23877 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 23878 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 23879 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 23880 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 23881 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 23882 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 23883 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 23884 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1532, + "step": 23885 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1652, + "step": 23886 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1542, + "step": 23887 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1684, + "step": 23888 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 23889 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1538, + "step": 23890 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 23891 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1718, + "step": 23892 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 23893 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 23894 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 23895 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1597, + "step": 23896 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 23897 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 23898 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 23899 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 23900 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1583, + "step": 23901 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1513, + "step": 23902 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1561, + "step": 23903 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 23904 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 23905 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 23906 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 23907 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 23908 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 23909 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 23910 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 23911 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1646, + "step": 23912 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 23913 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1675, + "step": 23914 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 23915 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 23916 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 23917 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 23918 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 23919 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 23920 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.146, + "step": 23921 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1947, + "step": 23922 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 23923 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 23924 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 23925 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 23926 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 23927 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 23928 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 23929 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 23930 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1503, + "step": 23931 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 23932 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 23933 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 23934 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 23935 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 23936 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 23937 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 23938 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 23939 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1942, + "step": 23940 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 23941 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 23942 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 23943 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 23944 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1735, + "step": 23945 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.1551, + "step": 23946 + }, + { + "epoch": 1.45, + "learning_rate": 2e-05, + "loss": 0.161, + "step": 23947 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 23948 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 23949 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1552, + "step": 23950 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 23951 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 23952 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 23953 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 23954 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 23955 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 23956 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 23957 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.155, + "step": 23958 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 23959 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 23960 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 23961 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 23962 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 23963 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 23964 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 23965 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 23966 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 23967 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 23968 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1632, + "step": 23969 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 23970 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1758, + "step": 23971 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 23972 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 23973 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.0867, + "step": 23974 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 23975 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1592, + "step": 23976 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 23977 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 23978 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1462, + "step": 23979 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 23980 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 23981 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 23982 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 23983 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 23984 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 23985 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 23986 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 23987 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 23988 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 23989 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 23990 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 23991 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.0936, + "step": 23992 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 23993 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 23994 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 23995 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 23996 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 23997 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 23998 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 23999 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 24000 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 24001 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1627, + "step": 24002 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 24003 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 24004 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 24005 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1476, + "step": 24006 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 24007 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 24008 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 24009 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 24010 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 24011 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 24012 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1694, + "step": 24013 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 24014 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1675, + "step": 24015 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 24016 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 24017 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 24018 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 24019 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 24020 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 24021 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1835, + "step": 24022 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 24023 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1712, + "step": 24024 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1761, + "step": 24025 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 24026 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 24027 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 24028 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 24029 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 24030 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 24031 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 24032 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 24033 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 24034 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 24035 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 24036 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1736, + "step": 24037 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 24038 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 24039 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 24040 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 24041 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 24042 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1665, + "step": 24043 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 24044 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 24045 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1595, + "step": 24046 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 24047 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1573, + "step": 24048 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 24049 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 24050 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1532, + "step": 24051 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 24052 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 24053 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 24054 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1689, + "step": 24055 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 24056 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.146, + "step": 24057 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 24058 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 24059 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 24060 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 24061 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 24062 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 24063 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 24064 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1525, + "step": 24065 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1721, + "step": 24066 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 24067 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 24068 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 24069 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 24070 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 24071 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1744, + "step": 24072 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 24073 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1465, + "step": 24074 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 24075 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 24076 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1597, + "step": 24077 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1624, + "step": 24078 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 24079 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1719, + "step": 24080 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 24081 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 24082 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 24083 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 24084 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1582, + "step": 24085 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1702, + "step": 24086 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 24087 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 24088 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 24089 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1699, + "step": 24090 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 24091 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1595, + "step": 24092 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 24093 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 24094 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 24095 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 24096 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 24097 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 24098 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 24099 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 24100 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 24101 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 24102 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1962, + "step": 24103 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 24104 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 24105 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 24106 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1693, + "step": 24107 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1642, + "step": 24108 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 24109 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 24110 + }, + { + "epoch": 1.46, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 24111 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 24112 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1582, + "step": 24113 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 24114 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 24115 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 24116 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 24117 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 24118 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 24119 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 24120 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 24121 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 24122 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 24123 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 24124 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 24125 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 24126 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 24127 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 24128 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 24129 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 24130 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 24131 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1884, + "step": 24132 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1686, + "step": 24133 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1665, + "step": 24134 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 24135 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 24136 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 24137 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1498, + "step": 24138 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1825, + "step": 24139 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 24140 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 24141 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 24142 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 24143 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 24144 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1542, + "step": 24145 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 24146 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 24147 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 24148 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 24149 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 24150 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 24151 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 24152 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 24153 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1625, + "step": 24154 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 24155 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1517, + "step": 24156 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 24157 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 24158 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 24159 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 24160 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1728, + "step": 24161 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 24162 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 24163 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 24164 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 24165 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1485, + "step": 24166 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 24167 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 24168 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.177, + "step": 24169 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 24170 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 24171 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 24172 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 24173 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 24174 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 24175 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1503, + "step": 24176 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 24177 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 24178 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 24179 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.175, + "step": 24180 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 24181 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 24182 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1663, + "step": 24183 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 24184 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 24185 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 24186 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1542, + "step": 24187 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 24188 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 24189 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 24190 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 24191 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 24192 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 24193 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 24194 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 24195 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 24196 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 24197 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 24198 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 24199 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 24200 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 24201 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 24202 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1634, + "step": 24203 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 24204 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 24205 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 24206 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 24207 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1573, + "step": 24208 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 24209 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1513, + "step": 24210 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1589, + "step": 24211 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 24212 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1605, + "step": 24213 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 24214 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 24215 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 24216 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 24217 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 24218 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 24219 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1723, + "step": 24220 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 24221 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 24222 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 24223 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1881, + "step": 24224 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1669, + "step": 24225 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1677, + "step": 24226 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 24227 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 24228 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.171, + "step": 24229 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 24230 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 24231 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.0903, + "step": 24232 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 24233 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 24234 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 24235 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1553, + "step": 24236 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 24237 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 24238 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 24239 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 24240 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 24241 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 24242 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 24243 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 24244 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 24245 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1714, + "step": 24246 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 24247 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 24248 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 24249 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 24250 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 24251 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 24252 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1579, + "step": 24253 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 24254 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1459, + "step": 24255 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1647, + "step": 24256 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 24257 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 24258 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 24259 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 24260 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 24261 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 24262 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 24263 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 24264 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 24265 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 24266 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1497, + "step": 24267 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 24268 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 24269 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 24270 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 24271 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 24272 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 24273 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.173, + "step": 24274 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 24275 + }, + { + "epoch": 1.47, + "learning_rate": 2e-05, + "loss": 0.0871, + "step": 24276 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 24277 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 24278 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 24279 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 24280 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 24281 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 24282 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.169, + "step": 24283 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 24284 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 24285 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1552, + "step": 24286 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1735, + "step": 24287 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 24288 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 24289 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 24290 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.146, + "step": 24291 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 24292 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 24293 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 24294 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1459, + "step": 24295 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 24296 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 24297 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 24298 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 24299 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 24300 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.161, + "step": 24301 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 24302 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 24303 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 24304 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 24305 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 24306 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 24307 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 24308 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 24309 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 24310 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 24311 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 24312 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 24313 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 24314 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1495, + "step": 24315 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 24316 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1485, + "step": 24317 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 24318 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 24319 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 24320 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1604, + "step": 24321 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 24322 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.186, + "step": 24323 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 24324 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 24325 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 24326 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 24327 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 24328 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 24329 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 24330 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 24331 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 24332 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1488, + "step": 24333 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 24334 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 24335 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 24336 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 24337 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 24338 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 24339 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.154, + "step": 24340 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 24341 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 24342 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1535, + "step": 24343 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 24344 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 24345 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 24346 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1637, + "step": 24347 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1793, + "step": 24348 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 24349 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1617, + "step": 24350 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 24351 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 24352 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1582, + "step": 24353 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 24354 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 24355 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 24356 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 24357 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 24358 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 24359 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 24360 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 24361 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1727, + "step": 24362 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 24363 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 24364 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 24365 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 24366 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 24367 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 24368 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1647, + "step": 24369 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 24370 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 24371 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 24372 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 24373 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 24374 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 24375 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 24376 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1555, + "step": 24377 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1645, + "step": 24378 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 24379 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1809, + "step": 24380 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1451, + "step": 24381 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 24382 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 24383 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.16, + "step": 24384 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 24385 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 24386 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 24387 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 24388 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 24389 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 24390 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1592, + "step": 24391 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1507, + "step": 24392 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 24393 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 24394 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 24395 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 24396 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1589, + "step": 24397 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 24398 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 24399 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 24400 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 24401 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 24402 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 24403 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 24404 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1442, + "step": 24405 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 24406 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1451, + "step": 24407 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 24408 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 24409 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1788, + "step": 24410 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 24411 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1503, + "step": 24412 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 24413 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 24414 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 24415 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 24416 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 24417 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1616, + "step": 24418 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 24419 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 24420 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 24421 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 24422 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 24423 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1656, + "step": 24424 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.164, + "step": 24425 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 24426 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1645, + "step": 24427 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.2167, + "step": 24428 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1528, + "step": 24429 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 24430 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 24431 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1488, + "step": 24432 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 24433 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 24434 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 24435 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 24436 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 24437 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 24438 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1982, + "step": 24439 + }, + { + "epoch": 1.48, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 24440 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1624, + "step": 24441 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 24442 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1871, + "step": 24443 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 24444 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 24445 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1698, + "step": 24446 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1656, + "step": 24447 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 24448 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 24449 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1728, + "step": 24450 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 24451 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1685, + "step": 24452 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 24453 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 24454 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1447, + "step": 24455 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 24456 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 24457 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 24458 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 24459 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 24460 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 24461 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 24462 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 24463 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1679, + "step": 24464 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1429, + "step": 24465 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 24466 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1513, + "step": 24467 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 24468 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.0901, + "step": 24469 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 24470 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 24471 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 24472 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 24473 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 24474 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.176, + "step": 24475 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 24476 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.144, + "step": 24477 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.2011, + "step": 24478 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.152, + "step": 24479 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 24480 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 24481 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 24482 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 24483 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 24484 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1519, + "step": 24485 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 24486 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 24487 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 24488 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 24489 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.0909, + "step": 24490 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1624, + "step": 24491 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 24492 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 24493 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 24494 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 24495 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 24496 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1694, + "step": 24497 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 24498 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1627, + "step": 24499 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 24500 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 24501 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1704, + "step": 24502 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1495, + "step": 24503 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 24504 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 24505 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 24506 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1557, + "step": 24507 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 24508 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 24509 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 24510 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 24511 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 24512 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 24513 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 24514 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 24515 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1548, + "step": 24516 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1649, + "step": 24517 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 24518 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 24519 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 24520 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 24521 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 24522 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 24523 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1813, + "step": 24524 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 24525 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 24526 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 24527 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 24528 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 24529 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 24530 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1525, + "step": 24531 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 24532 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 24533 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 24534 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 24535 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 24536 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 24537 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 24538 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 24539 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1589, + "step": 24540 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 24541 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 24542 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 24543 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 24544 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 24545 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 24546 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 24547 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 24548 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 24549 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 24550 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 24551 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 24552 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 24553 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 24554 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 24555 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 24556 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 24557 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 24558 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 24559 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1735, + "step": 24560 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1505, + "step": 24561 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 24562 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1648, + "step": 24563 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 24564 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 24565 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 24566 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1927, + "step": 24567 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 24568 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1881, + "step": 24569 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 24570 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 24571 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 24572 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 24573 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 24574 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 24575 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 24576 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 24577 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 24578 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 24579 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1578, + "step": 24580 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 24581 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1573, + "step": 24582 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 24583 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 24584 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 24585 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 24586 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 24587 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1834, + "step": 24588 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 24589 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1602, + "step": 24590 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 24591 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 24592 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 24593 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.16, + "step": 24594 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 24595 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 24596 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1665, + "step": 24597 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 24598 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 24599 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 24600 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 24601 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 24602 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 24603 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1447, + "step": 24604 + }, + { + "epoch": 1.49, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 24605 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 24606 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 24607 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1596, + "step": 24608 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 24609 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 24610 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1478, + "step": 24611 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 24612 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 24613 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 24614 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 24615 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 24616 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1582, + "step": 24617 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1784, + "step": 24618 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 24619 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 24620 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 24621 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 24622 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 24623 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 24624 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1846, + "step": 24625 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1663, + "step": 24626 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 24627 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 24628 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 24629 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 24630 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 24631 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 24632 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 24633 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 24634 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 24635 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1733, + "step": 24636 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 24637 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 24638 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.171, + "step": 24639 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 24640 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 24641 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 24642 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1626, + "step": 24643 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.174, + "step": 24644 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1616, + "step": 24645 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 24646 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 24647 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 24648 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 24649 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1557, + "step": 24650 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 24651 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 24652 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 24653 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 24654 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 24655 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1644, + "step": 24656 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 24657 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 24658 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 24659 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1725, + "step": 24660 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1515, + "step": 24661 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 24662 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 24663 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 24664 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1679, + "step": 24665 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 24666 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 24667 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 24668 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 24669 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 24670 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 24671 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 24672 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 24673 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1453, + "step": 24674 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 24675 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 24676 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 24677 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 24678 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 24679 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 24680 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 24681 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 24682 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 24683 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1548, + "step": 24684 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 24685 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 24686 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 24687 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 24688 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 24689 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 24690 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1604, + "step": 24691 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 24692 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1585, + "step": 24693 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1798, + "step": 24694 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 24695 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1876, + "step": 24696 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 24697 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 24698 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 24699 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 24700 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 24701 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.0882, + "step": 24702 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 24703 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1764, + "step": 24704 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 24705 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 24706 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1546, + "step": 24707 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 24708 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 24709 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1585, + "step": 24710 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1641, + "step": 24711 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1806, + "step": 24712 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 24713 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 24714 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 24715 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 24716 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 24717 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 24718 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 24719 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1668, + "step": 24720 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 24721 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 24722 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 24723 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 24724 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 24725 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 24726 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 24727 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 24728 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 24729 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 24730 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 24731 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 24732 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 24733 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1652, + "step": 24734 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 24735 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 24736 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 24737 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 24738 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1684, + "step": 24739 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1564, + "step": 24740 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 24741 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.164, + "step": 24742 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 24743 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1685, + "step": 24744 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1564, + "step": 24745 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 24746 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1531, + "step": 24747 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 24748 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 24749 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 24750 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 24751 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1546, + "step": 24752 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 24753 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1688, + "step": 24754 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 24755 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.0854, + "step": 24756 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 24757 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 24758 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 24759 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 24760 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.171, + "step": 24761 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1855, + "step": 24762 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1443, + "step": 24763 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 24764 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 24765 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 24766 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 24767 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 24768 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 24769 + }, + { + "epoch": 1.5, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 24770 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 24771 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 24772 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 24773 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1503, + "step": 24774 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 24775 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1519, + "step": 24776 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 24777 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 24778 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 24779 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 24780 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 24781 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 24782 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 24783 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 24784 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 24785 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 24786 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1488, + "step": 24787 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 24788 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1706, + "step": 24789 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 24790 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1478, + "step": 24791 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 24792 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1505, + "step": 24793 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 24794 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 24795 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 24796 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 24797 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 24798 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 24799 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 24800 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 24801 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1616, + "step": 24802 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1698, + "step": 24803 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 24804 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 24805 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1603, + "step": 24806 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 24807 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 24808 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 24809 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1451, + "step": 24810 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 24811 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1669, + "step": 24812 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 24813 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 24814 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 24815 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 24816 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 24817 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 24818 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 24819 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1901, + "step": 24820 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 24821 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 24822 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 24823 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 24824 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 24825 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1553, + "step": 24826 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 24827 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 24828 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 24829 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1531, + "step": 24830 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 24831 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 24832 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 24833 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1583, + "step": 24834 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1637, + "step": 24835 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 24836 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1579, + "step": 24837 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.145, + "step": 24838 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 24839 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1828, + "step": 24840 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 24841 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 24842 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.178, + "step": 24843 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 24844 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1869, + "step": 24845 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.147, + "step": 24846 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1734, + "step": 24847 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 24848 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 24849 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 24850 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1776, + "step": 24851 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 24852 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 24853 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 24854 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 24855 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 24856 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 24857 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 24858 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 24859 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1806, + "step": 24860 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 24861 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1605, + "step": 24862 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1721, + "step": 24863 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 24864 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 24865 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.2128, + "step": 24866 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 24867 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 24868 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 24869 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 24870 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 24871 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 24872 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1554, + "step": 24873 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 24874 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1745, + "step": 24875 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 24876 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1542, + "step": 24877 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 24878 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 24879 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1451, + "step": 24880 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.173, + "step": 24881 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 24882 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 24883 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 24884 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 24885 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 24886 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 24887 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 24888 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 24889 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 24890 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1676, + "step": 24891 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 24892 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.146, + "step": 24893 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 24894 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 24895 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1549, + "step": 24896 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 24897 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 24898 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 24899 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 24900 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 24901 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 24902 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 24903 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 24904 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 24905 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 24906 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1732, + "step": 24907 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 24908 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1623, + "step": 24909 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 24910 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 24911 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 24912 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 24913 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 24914 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 24915 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 24916 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 24917 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 24918 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.148, + "step": 24919 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 24920 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 24921 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 24922 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 24923 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1805, + "step": 24924 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 24925 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 24926 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 24927 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 24928 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 24929 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 24930 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1558, + "step": 24931 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 24932 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 24933 + }, + { + "epoch": 1.51, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 24934 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 24935 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1503, + "step": 24936 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 24937 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 24938 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1626, + "step": 24939 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 24940 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 24941 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1593, + "step": 24942 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 24943 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 24944 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 24945 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 24946 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1503, + "step": 24947 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 24948 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 24949 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 24950 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 24951 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 24952 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 24953 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 24954 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 24955 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1603, + "step": 24956 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 24957 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 24958 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 24959 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 24960 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 24961 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 24962 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 24963 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 24964 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1692, + "step": 24965 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 24966 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1647, + "step": 24967 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1673, + "step": 24968 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 24969 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 24970 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 24971 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 24972 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.2026, + "step": 24973 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 24974 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1639, + "step": 24975 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1743, + "step": 24976 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 24977 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 24978 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 24979 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 24980 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 24981 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1644, + "step": 24982 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 24983 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 24984 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 24985 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 24986 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1624, + "step": 24987 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1599, + "step": 24988 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 24989 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1596, + "step": 24990 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.161, + "step": 24991 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 24992 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 24993 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 24994 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 24995 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 24996 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 24997 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 24998 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 24999 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 25000 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1867, + "step": 25001 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 25002 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 25003 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 25004 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 25005 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 25006 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 25007 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1555, + "step": 25008 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 25009 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.144, + "step": 25010 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.179, + "step": 25011 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 25012 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.0877, + "step": 25013 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 25014 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 25015 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 25016 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 25017 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 25018 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1505, + "step": 25019 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 25020 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 25021 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.172, + "step": 25022 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 25023 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1641, + "step": 25024 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 25025 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 25026 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 25027 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1561, + "step": 25028 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 25029 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 25030 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 25031 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 25032 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 25033 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 25034 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1535, + "step": 25035 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 25036 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 25037 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 25038 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 25039 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 25040 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 25041 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 25042 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 25043 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 25044 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 25045 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1608, + "step": 25046 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 25047 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 25048 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 25049 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 25050 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 25051 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 25052 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 25053 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 25054 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 25055 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 25056 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 25057 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 25058 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1825, + "step": 25059 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1586, + "step": 25060 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 25061 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1709, + "step": 25062 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 25063 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 25064 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1507, + "step": 25065 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 25066 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1677, + "step": 25067 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 25068 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 25069 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1666, + "step": 25070 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 25071 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 25072 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 25073 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 25074 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1476, + "step": 25075 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 25076 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 25077 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 25078 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 25079 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1849, + "step": 25080 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 25081 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 25082 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 25083 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 25084 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 25085 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 25086 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1521, + "step": 25087 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 25088 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1516, + "step": 25089 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 25090 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 25091 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 25092 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 25093 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1902, + "step": 25094 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1541, + "step": 25095 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 25096 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1651, + "step": 25097 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1709, + "step": 25098 + }, + { + "epoch": 1.52, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 25099 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 25100 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 25101 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 25102 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 25103 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 25104 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 25105 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 25106 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 25107 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 25108 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 25109 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1695, + "step": 25110 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1469, + "step": 25111 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1617, + "step": 25112 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1804, + "step": 25113 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 25114 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 25115 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 25116 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 25117 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 25118 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1612, + "step": 25119 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1573, + "step": 25120 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1632, + "step": 25121 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1765, + "step": 25122 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 25123 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 25124 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 25125 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1795, + "step": 25126 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 25127 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 25128 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 25129 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1577, + "step": 25130 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 25131 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 25132 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1643, + "step": 25133 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 25134 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 25135 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 25136 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 25137 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1753, + "step": 25138 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1453, + "step": 25139 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 25140 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1918, + "step": 25141 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 25142 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1541, + "step": 25143 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1633, + "step": 25144 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1469, + "step": 25145 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 25146 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 25147 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 25148 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 25149 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 25150 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 25151 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 25152 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 25153 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 25154 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1682, + "step": 25155 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 25156 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 25157 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1728, + "step": 25158 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 25159 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.148, + "step": 25160 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 25161 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1654, + "step": 25162 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.0875, + "step": 25163 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 25164 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 25165 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1533, + "step": 25166 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 25167 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 25168 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 25169 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1796, + "step": 25170 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 25171 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 25172 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 25173 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 25174 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 25175 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 25176 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 25177 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 25178 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 25179 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 25180 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 25181 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1685, + "step": 25182 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 25183 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 25184 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 25185 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 25186 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 25187 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 25188 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 25189 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 25190 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1695, + "step": 25191 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 25192 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 25193 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 25194 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 25195 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 25196 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 25197 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1836, + "step": 25198 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 25199 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 25200 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 25201 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 25202 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1894, + "step": 25203 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 25204 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 25205 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 25206 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 25207 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 25208 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1542, + "step": 25209 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1442, + "step": 25210 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 25211 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 25212 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1655, + "step": 25213 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 25214 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1704, + "step": 25215 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 25216 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1728, + "step": 25217 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 25218 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 25219 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 25220 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 25221 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 25222 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 25223 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 25224 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1586, + "step": 25225 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.0903, + "step": 25226 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1623, + "step": 25227 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 25228 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 25229 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 25230 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1747, + "step": 25231 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 25232 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1525, + "step": 25233 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 25234 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 25235 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 25236 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 25237 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 25238 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 25239 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 25240 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 25241 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1822, + "step": 25242 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 25243 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1693, + "step": 25244 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 25245 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1809, + "step": 25246 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1641, + "step": 25247 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 25248 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 25249 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 25250 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 25251 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 25252 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 25253 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1712, + "step": 25254 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 25255 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 25256 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1621, + "step": 25257 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1561, + "step": 25258 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 25259 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 25260 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 25261 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 25262 + }, + { + "epoch": 1.53, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 25263 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 25264 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 25265 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 25266 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 25267 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1553, + "step": 25268 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 25269 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 25270 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 25271 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.16, + "step": 25272 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1788, + "step": 25273 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 25274 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 25275 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1997, + "step": 25276 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 25277 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 25278 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 25279 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 25280 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 25281 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 25282 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1517, + "step": 25283 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 25284 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 25285 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 25286 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 25287 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 25288 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1727, + "step": 25289 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 25290 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 25291 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 25292 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 25293 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1722, + "step": 25294 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 25295 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 25296 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1727, + "step": 25297 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1595, + "step": 25298 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1583, + "step": 25299 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 25300 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 25301 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1726, + "step": 25302 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1453, + "step": 25303 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 25304 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 25305 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 25306 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 25307 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 25308 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.2299, + "step": 25309 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 25310 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1607, + "step": 25311 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 25312 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 25313 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 25314 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 25315 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 25316 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 25317 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 25318 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1672, + "step": 25319 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 25320 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 25321 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 25322 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 25323 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 25324 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 25325 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.176, + "step": 25326 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 25327 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 25328 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 25329 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 25330 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1903, + "step": 25331 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 25332 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 25333 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1787, + "step": 25334 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 25335 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 25336 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 25337 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 25338 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1767, + "step": 25339 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 25340 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 25341 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 25342 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 25343 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 25344 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 25345 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 25346 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 25347 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1641, + "step": 25348 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 25349 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 25350 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1617, + "step": 25351 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 25352 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 25353 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1822, + "step": 25354 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 25355 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1538, + "step": 25356 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.0853, + "step": 25357 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 25358 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 25359 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 25360 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 25361 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 25362 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 25363 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 25364 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 25365 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1975, + "step": 25366 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1451, + "step": 25367 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 25368 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 25369 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1608, + "step": 25370 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 25371 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 25372 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 25373 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 25374 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 25375 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1526, + "step": 25376 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1518, + "step": 25377 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 25378 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 25379 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 25380 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1867, + "step": 25381 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 25382 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 25383 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1856, + "step": 25384 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 25385 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1573, + "step": 25386 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1459, + "step": 25387 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1661, + "step": 25388 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1599, + "step": 25389 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 25390 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 25391 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 25392 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1602, + "step": 25393 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 25394 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 25395 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 25396 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1643, + "step": 25397 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 25398 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1782, + "step": 25399 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 25400 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 25401 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 25402 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 25403 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 25404 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 25405 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 25406 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 25407 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1781, + "step": 25408 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 25409 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 25410 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 25411 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 25412 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 25413 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 25414 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 25415 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1573, + "step": 25416 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 25417 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 25418 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1566, + "step": 25419 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 25420 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 25421 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 25422 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 25423 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 25424 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 25425 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 25426 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 25427 + }, + { + "epoch": 1.54, + "learning_rate": 2e-05, + "loss": 0.1456, + "step": 25428 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1781, + "step": 25429 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 25430 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1888, + "step": 25431 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.228, + "step": 25432 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 25433 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 25434 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 25435 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 25436 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1633, + "step": 25437 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 25438 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 25439 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 25440 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 25441 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 25442 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 25443 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 25444 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.161, + "step": 25445 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 25446 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 25447 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 25448 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1654, + "step": 25449 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 25450 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1683, + "step": 25451 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 25452 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 25453 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 25454 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 25455 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 25456 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1668, + "step": 25457 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 25458 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1562, + "step": 25459 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 25460 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 25461 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 25462 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1498, + "step": 25463 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 25464 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 25465 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 25466 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 25467 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 25468 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.168, + "step": 25469 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1551, + "step": 25470 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1515, + "step": 25471 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 25472 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 25473 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 25474 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1949, + "step": 25475 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1933, + "step": 25476 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 25477 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 25478 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 25479 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1603, + "step": 25480 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1527, + "step": 25481 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 25482 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 25483 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 25484 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 25485 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 25486 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 25487 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 25488 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 25489 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 25490 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 25491 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 25492 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 25493 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 25494 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1747, + "step": 25495 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 25496 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 25497 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 25498 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1826, + "step": 25499 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 25500 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 25501 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 25502 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 25503 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 25504 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 25505 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1659, + "step": 25506 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1601, + "step": 25507 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1749, + "step": 25508 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 25509 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 25510 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 25511 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 25512 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1937, + "step": 25513 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1666, + "step": 25514 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 25515 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 25516 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 25517 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 25518 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 25519 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1741, + "step": 25520 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 25521 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 25522 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1772, + "step": 25523 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1648, + "step": 25524 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.171, + "step": 25525 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 25526 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1606, + "step": 25527 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 25528 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 25529 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 25530 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 25531 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.0932, + "step": 25532 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 25533 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 25534 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 25535 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 25536 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 25537 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1599, + "step": 25538 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1641, + "step": 25539 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 25540 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 25541 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1892, + "step": 25542 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 25543 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 25544 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 25545 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 25546 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 25547 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 25548 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1485, + "step": 25549 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 25550 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 25551 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1936, + "step": 25552 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.0777, + "step": 25553 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1533, + "step": 25554 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 25555 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 25556 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 25557 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 25558 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 25559 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 25560 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1699, + "step": 25561 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 25562 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 25563 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 25564 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 25565 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 25566 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1753, + "step": 25567 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 25568 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 25569 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 25570 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 25571 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 25572 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 25573 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 25574 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.159, + "step": 25575 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 25576 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 25577 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 25578 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 25579 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 25580 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 25581 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 25582 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 25583 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 25584 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 25585 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.0828, + "step": 25586 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 25587 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 25588 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1639, + "step": 25589 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1533, + "step": 25590 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 25591 + }, + { + "epoch": 1.55, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 25592 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 25593 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 25594 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 25595 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1507, + "step": 25596 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 25597 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 25598 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 25599 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 25600 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 25601 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 25602 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 25603 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 25604 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 25605 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 25606 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 25607 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 25608 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 25609 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.214, + "step": 25610 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 25611 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 25612 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 25613 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 25614 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 25615 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.146, + "step": 25616 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1758, + "step": 25617 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 25618 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 25619 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 25620 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 25621 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 25622 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1796, + "step": 25623 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 25624 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1526, + "step": 25625 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 25626 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1645, + "step": 25627 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 25628 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 25629 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 25630 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 25631 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 25632 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.2065, + "step": 25633 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 25634 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 25635 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 25636 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 25637 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1456, + "step": 25638 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 25639 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 25640 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 25641 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1642, + "step": 25642 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 25643 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 25644 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 25645 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 25646 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 25647 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 25648 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 25649 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 25650 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 25651 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 25652 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 25653 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 25654 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 25655 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 25656 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 25657 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 25658 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 25659 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.185, + "step": 25660 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 25661 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 25662 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 25663 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 25664 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 25665 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1607, + "step": 25666 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 25667 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 25668 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1904, + "step": 25669 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 25670 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 25671 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 25672 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1573, + "step": 25673 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 25674 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 25675 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1608, + "step": 25676 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 25677 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 25678 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1642, + "step": 25679 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 25680 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 25681 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 25682 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 25683 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1657, + "step": 25684 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 25685 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 25686 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1597, + "step": 25687 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 25688 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.173, + "step": 25689 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1576, + "step": 25690 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1583, + "step": 25691 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 25692 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1443, + "step": 25693 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1542, + "step": 25694 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 25695 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 25696 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 25697 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 25698 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 25699 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 25700 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.174, + "step": 25701 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 25702 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 25703 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1949, + "step": 25704 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 25705 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 25706 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1459, + "step": 25707 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1629, + "step": 25708 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 25709 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 25710 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 25711 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 25712 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1718, + "step": 25713 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 25714 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 25715 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 25716 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 25717 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 25718 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 25719 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 25720 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 25721 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 25722 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.087, + "step": 25723 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 25724 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1758, + "step": 25725 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1727, + "step": 25726 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1546, + "step": 25727 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 25728 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 25729 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 25730 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 25731 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 25732 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 25733 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1689, + "step": 25734 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 25735 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 25736 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 25737 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 25738 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 25739 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 25740 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1729, + "step": 25741 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1641, + "step": 25742 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 25743 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 25744 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 25745 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 25746 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 25747 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 25748 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 25749 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 25750 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 25751 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 25752 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 25753 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 25754 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 25755 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 25756 + }, + { + "epoch": 1.56, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 25757 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 25758 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1658, + "step": 25759 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 25760 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1746, + "step": 25761 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 25762 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1735, + "step": 25763 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1519, + "step": 25764 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 25765 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 25766 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 25767 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 25768 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 25769 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 25770 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 25771 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 25772 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.156, + "step": 25773 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1891, + "step": 25774 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 25775 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 25776 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 25777 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.168, + "step": 25778 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 25779 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 25780 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 25781 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.182, + "step": 25782 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 25783 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1443, + "step": 25784 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 25785 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 25786 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 25787 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.196, + "step": 25788 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 25789 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 25790 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 25791 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 25792 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.171, + "step": 25793 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1955, + "step": 25794 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 25795 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1549, + "step": 25796 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 25797 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 25798 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.163, + "step": 25799 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1519, + "step": 25800 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 25801 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 25802 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 25803 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1595, + "step": 25804 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 25805 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 25806 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 25807 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 25808 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1601, + "step": 25809 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 25810 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 25811 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1911, + "step": 25812 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 25813 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 25814 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 25815 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1599, + "step": 25816 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 25817 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 25818 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1741, + "step": 25819 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 25820 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1519, + "step": 25821 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.156, + "step": 25822 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 25823 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 25824 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 25825 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1778, + "step": 25826 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 25827 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1469, + "step": 25828 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1631, + "step": 25829 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 25830 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 25831 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 25832 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 25833 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 25834 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 25835 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 25836 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1645, + "step": 25837 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 25838 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 25839 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 25840 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 25841 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 25842 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1507, + "step": 25843 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1657, + "step": 25844 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 25845 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 25846 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 25847 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 25848 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1804, + "step": 25849 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 25850 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 25851 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 25852 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 25853 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 25854 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 25855 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 25856 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 25857 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 25858 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 25859 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 25860 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1588, + "step": 25861 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1531, + "step": 25862 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 25863 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.146, + "step": 25864 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 25865 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 25866 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 25867 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1515, + "step": 25868 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1465, + "step": 25869 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 25870 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 25871 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 25872 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 25873 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 25874 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 25875 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 25876 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 25877 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1884, + "step": 25878 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 25879 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1708, + "step": 25880 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 25881 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 25882 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.2223, + "step": 25883 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 25884 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1488, + "step": 25885 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 25886 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 25887 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 25888 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1664, + "step": 25889 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 25890 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 25891 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 25892 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1519, + "step": 25893 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 25894 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 25895 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 25896 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 25897 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 25898 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 25899 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 25900 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 25901 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.187, + "step": 25902 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 25903 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 25904 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 25905 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 25906 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 25907 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 25908 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 25909 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 25910 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 25911 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1653, + "step": 25912 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1497, + "step": 25913 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 25914 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.162, + "step": 25915 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 25916 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1733, + "step": 25917 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 25918 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 25919 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 25920 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 25921 + }, + { + "epoch": 1.57, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 25922 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 25923 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 25924 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 25925 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 25926 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 25927 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 25928 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1755, + "step": 25929 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 25930 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 25931 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 25932 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.211, + "step": 25933 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1815, + "step": 25934 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 25935 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 25936 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 25937 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1521, + "step": 25938 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 25939 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 25940 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.2077, + "step": 25941 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 25942 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 25943 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1712, + "step": 25944 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 25945 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 25946 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 25947 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 25948 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 25949 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 25950 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 25951 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 25952 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 25953 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 25954 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 25955 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 25956 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 25957 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 25958 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1754, + "step": 25959 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 25960 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1647, + "step": 25961 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 25962 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1725, + "step": 25963 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 25964 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 25965 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 25966 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 25967 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1526, + "step": 25968 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 25969 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 25970 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 25971 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 25972 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 25973 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 25974 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 25975 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.158, + "step": 25976 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 25977 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1727, + "step": 25978 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 25979 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 25980 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 25981 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 25982 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 25983 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1731, + "step": 25984 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 25985 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1989, + "step": 25986 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 25987 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 25988 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 25989 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 25990 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 25991 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 25992 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 25993 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1661, + "step": 25994 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 25995 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 25996 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 25997 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 25998 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1443, + "step": 25999 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 26000 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 26001 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1499, + "step": 26002 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 26003 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 26004 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 26005 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.0872, + "step": 26006 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 26007 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 26008 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 26009 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 26010 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 26011 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 26012 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 26013 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 26014 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 26015 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 26016 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 26017 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 26018 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 26019 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 26020 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 26021 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 26022 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1795, + "step": 26023 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.0873, + "step": 26024 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 26025 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 26026 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 26027 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 26028 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1603, + "step": 26029 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.144, + "step": 26030 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 26031 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 26032 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1546, + "step": 26033 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 26034 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 26035 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1757, + "step": 26036 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 26037 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.162, + "step": 26038 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1503, + "step": 26039 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 26040 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 26041 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 26042 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 26043 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 26044 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 26045 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 26046 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 26047 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1526, + "step": 26048 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 26049 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 26050 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 26051 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 26052 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 26053 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 26054 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 26055 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 26056 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 26057 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 26058 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1683, + "step": 26059 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1564, + "step": 26060 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 26061 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 26062 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 26063 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1465, + "step": 26064 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 26065 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 26066 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 26067 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 26068 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 26069 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1607, + "step": 26070 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 26071 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 26072 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 26073 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 26074 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 26075 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 26076 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 26077 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 26078 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 26079 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 26080 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 26081 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 26082 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1528, + "step": 26083 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 26084 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 26085 + }, + { + "epoch": 1.58, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 26086 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 26087 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 26088 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 26089 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 26090 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.0918, + "step": 26091 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 26092 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 26093 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 26094 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 26095 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.2291, + "step": 26096 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 26097 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 26098 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 26099 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 26100 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1773, + "step": 26101 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 26102 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1521, + "step": 26103 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1937, + "step": 26104 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 26105 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 26106 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 26107 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 26108 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 26109 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 26110 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 26111 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 26112 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 26113 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 26114 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1503, + "step": 26115 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1736, + "step": 26116 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 26117 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 26118 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 26119 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 26120 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 26121 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.0785, + "step": 26122 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 26123 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1583, + "step": 26124 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.155, + "step": 26125 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.16, + "step": 26126 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 26127 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 26128 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1669, + "step": 26129 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 26130 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 26131 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 26132 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 26133 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1505, + "step": 26134 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 26135 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 26136 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 26137 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1746, + "step": 26138 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 26139 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1623, + "step": 26140 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 26141 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 26142 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1542, + "step": 26143 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 26144 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1534, + "step": 26145 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.0865, + "step": 26146 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 26147 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.0914, + "step": 26148 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 26149 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 26150 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 26151 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1726, + "step": 26152 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 26153 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1562, + "step": 26154 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.169, + "step": 26155 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1786, + "step": 26156 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 26157 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1693, + "step": 26158 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1525, + "step": 26159 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 26160 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1709, + "step": 26161 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 26162 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 26163 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 26164 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1697, + "step": 26165 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 26166 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 26167 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1737, + "step": 26168 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 26169 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 26170 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 26171 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1746, + "step": 26172 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 26173 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 26174 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 26175 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 26176 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 26177 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 26178 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1478, + "step": 26179 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 26180 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 26181 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 26182 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 26183 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 26184 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 26185 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 26186 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 26187 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1566, + "step": 26188 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 26189 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 26190 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 26191 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 26192 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1531, + "step": 26193 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1447, + "step": 26194 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1967, + "step": 26195 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1622, + "step": 26196 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 26197 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 26198 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 26199 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 26200 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 26201 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 26202 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1697, + "step": 26203 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 26204 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 26205 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 26206 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 26207 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1691, + "step": 26208 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 26209 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 26210 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 26211 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1635, + "step": 26212 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 26213 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 26214 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1681, + "step": 26215 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.165, + "step": 26216 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 26217 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 26218 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1693, + "step": 26219 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 26220 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1704, + "step": 26221 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 26222 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1526, + "step": 26223 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 26224 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1758, + "step": 26225 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 26226 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1561, + "step": 26227 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1627, + "step": 26228 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 26229 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 26230 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 26231 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1985, + "step": 26232 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 26233 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 26234 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1663, + "step": 26235 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 26236 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 26237 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 26238 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 26239 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 26240 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 26241 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 26242 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 26243 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 26244 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1519, + "step": 26245 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 26246 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1622, + "step": 26247 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 26248 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 26249 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 26250 + }, + { + "epoch": 1.59, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 26251 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1563, + "step": 26252 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 26253 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 26254 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 26255 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 26256 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1919, + "step": 26257 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 26258 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 26259 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 26260 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 26261 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 26262 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1793, + "step": 26263 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 26264 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 26265 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 26266 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 26267 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 26268 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1583, + "step": 26269 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 26270 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 26271 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1839, + "step": 26272 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 26273 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.0903, + "step": 26274 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 26275 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 26276 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1665, + "step": 26277 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1826, + "step": 26278 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1631, + "step": 26279 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 26280 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 26281 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 26282 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.2044, + "step": 26283 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.167, + "step": 26284 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.0853, + "step": 26285 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 26286 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 26287 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 26288 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 26289 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 26290 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 26291 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 26292 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 26293 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 26294 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1891, + "step": 26295 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 26296 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 26297 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 26298 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 26299 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 26300 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 26301 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 26302 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 26303 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 26304 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 26305 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 26306 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 26307 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1538, + "step": 26308 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 26309 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1567, + "step": 26310 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1763, + "step": 26311 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 26312 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1527, + "step": 26313 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 26314 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 26315 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 26316 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 26317 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 26318 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 26319 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 26320 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 26321 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 26322 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 26323 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 26324 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 26325 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.167, + "step": 26326 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 26327 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 26328 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 26329 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1558, + "step": 26330 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 26331 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 26332 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 26333 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 26334 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 26335 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 26336 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 26337 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 26338 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1534, + "step": 26339 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 26340 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 26341 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 26342 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 26343 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 26344 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1465, + "step": 26345 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1577, + "step": 26346 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 26347 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 26348 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 26349 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 26350 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 26351 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 26352 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 26353 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 26354 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 26355 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1628, + "step": 26356 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1629, + "step": 26357 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 26358 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 26359 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 26360 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 26361 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 26362 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1499, + "step": 26363 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 26364 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1459, + "step": 26365 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.177, + "step": 26366 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.0779, + "step": 26367 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 26368 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 26369 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 26370 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 26371 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1739, + "step": 26372 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1566, + "step": 26373 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 26374 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 26375 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1689, + "step": 26376 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 26377 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 26378 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 26379 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 26380 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 26381 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 26382 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 26383 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 26384 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 26385 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1691, + "step": 26386 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 26387 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 26388 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 26389 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1589, + "step": 26390 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.0782, + "step": 26391 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.194, + "step": 26392 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 26393 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 26394 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 26395 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 26396 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 26397 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1447, + "step": 26398 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 26399 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 26400 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 26401 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 26402 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1983, + "step": 26403 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 26404 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 26405 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 26406 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 26407 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 26408 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 26409 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1573, + "step": 26410 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 26411 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 26412 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 26413 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 26414 + }, + { + "epoch": 1.6, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 26415 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1614, + "step": 26416 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 26417 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1613, + "step": 26418 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 26419 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1563, + "step": 26420 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 26421 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 26422 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 26423 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 26424 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 26425 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 26426 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 26427 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 26428 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 26429 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1719, + "step": 26430 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 26431 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 26432 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 26433 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 26434 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 26435 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1604, + "step": 26436 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1671, + "step": 26437 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 26438 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 26439 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 26440 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 26441 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1791, + "step": 26442 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 26443 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 26444 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 26445 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 26446 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1582, + "step": 26447 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 26448 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 26449 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1794, + "step": 26450 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 26451 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 26452 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 26453 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1831, + "step": 26454 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 26455 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1755, + "step": 26456 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 26457 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 26458 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 26459 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 26460 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.156, + "step": 26461 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 26462 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 26463 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1975, + "step": 26464 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 26465 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 26466 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 26467 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1533, + "step": 26468 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.17, + "step": 26469 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 26470 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1591, + "step": 26471 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1685, + "step": 26472 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 26473 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1447, + "step": 26474 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 26475 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 26476 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 26477 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 26478 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 26479 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 26480 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 26481 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 26482 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1562, + "step": 26483 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 26484 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1605, + "step": 26485 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 26486 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1684, + "step": 26487 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.145, + "step": 26488 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 26489 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1599, + "step": 26490 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 26491 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 26492 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 26493 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 26494 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 26495 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 26496 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 26497 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1822, + "step": 26498 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 26499 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1456, + "step": 26500 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1605, + "step": 26501 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 26502 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1729, + "step": 26503 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 26504 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1577, + "step": 26505 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 26506 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 26507 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1601, + "step": 26508 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 26509 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 26510 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 26511 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1809, + "step": 26512 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.2084, + "step": 26513 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 26514 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 26515 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 26516 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 26517 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1608, + "step": 26518 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 26519 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 26520 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 26521 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1751, + "step": 26522 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1554, + "step": 26523 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 26524 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 26525 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 26526 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 26527 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1802, + "step": 26528 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 26529 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 26530 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 26531 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 26532 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 26533 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 26534 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 26535 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 26536 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 26537 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 26538 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1528, + "step": 26539 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 26540 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 26541 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 26542 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 26543 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 26544 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 26545 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 26546 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 26547 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1465, + "step": 26548 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 26549 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 26550 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1585, + "step": 26551 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 26552 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 26553 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 26554 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1688, + "step": 26555 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 26556 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 26557 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 26558 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 26559 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.2225, + "step": 26560 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1794, + "step": 26561 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.155, + "step": 26562 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 26563 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 26564 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 26565 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 26566 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 26567 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 26568 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 26569 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 26570 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1597, + "step": 26571 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 26572 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 26573 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 26574 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 26575 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1638, + "step": 26576 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 26577 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 26578 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 26579 + }, + { + "epoch": 1.61, + "learning_rate": 2e-05, + "loss": 0.1702, + "step": 26580 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 26581 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 26582 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 26583 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 26584 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1679, + "step": 26585 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 26586 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1465, + "step": 26587 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 26588 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 26589 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 26590 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1488, + "step": 26591 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 26592 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 26593 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 26594 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1488, + "step": 26595 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1526, + "step": 26596 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1689, + "step": 26597 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 26598 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 26599 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 26600 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1453, + "step": 26601 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 26602 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.2164, + "step": 26603 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1443, + "step": 26604 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 26605 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 26606 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1846, + "step": 26607 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 26608 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1462, + "step": 26609 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 26610 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 26611 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1585, + "step": 26612 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.159, + "step": 26613 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 26614 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1443, + "step": 26615 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 26616 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.154, + "step": 26617 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 26618 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 26619 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 26620 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1562, + "step": 26621 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 26622 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 26623 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 26624 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1762, + "step": 26625 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 26626 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 26627 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1598, + "step": 26628 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 26629 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 26630 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 26631 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 26632 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 26633 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 26634 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 26635 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 26636 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1697, + "step": 26637 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 26638 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 26639 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 26640 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 26641 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 26642 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 26643 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 26644 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 26645 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 26646 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.198, + "step": 26647 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 26648 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 26649 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 26650 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 26651 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 26652 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 26653 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 26654 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 26655 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 26656 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1442, + "step": 26657 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 26658 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 26659 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 26660 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 26661 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 26662 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1756, + "step": 26663 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 26664 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 26665 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 26666 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 26667 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 26668 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 26669 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.16, + "step": 26670 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 26671 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 26672 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 26673 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 26674 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 26675 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 26676 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 26677 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 26678 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 26679 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 26680 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 26681 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 26682 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 26683 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 26684 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 26685 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 26686 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 26687 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 26688 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 26689 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 26690 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 26691 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 26692 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 26693 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 26694 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 26695 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.147, + "step": 26696 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 26697 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 26698 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 26699 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1746, + "step": 26700 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 26701 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 26702 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 26703 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 26704 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1646, + "step": 26705 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 26706 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 26707 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 26708 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.183, + "step": 26709 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.152, + "step": 26710 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1521, + "step": 26711 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 26712 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 26713 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 26714 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.0829, + "step": 26715 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 26716 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 26717 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 26718 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 26719 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1611, + "step": 26720 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 26721 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 26722 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 26723 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 26724 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 26725 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 26726 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 26727 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 26728 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 26729 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1655, + "step": 26730 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 26731 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 26732 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 26733 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 26734 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 26735 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1526, + "step": 26736 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 26737 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1447, + "step": 26738 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1549, + "step": 26739 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1541, + "step": 26740 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 26741 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 26742 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1605, + "step": 26743 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.1675, + "step": 26744 + }, + { + "epoch": 1.62, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 26745 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 26746 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 26747 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.17, + "step": 26748 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 26749 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 26750 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 26751 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 26752 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 26753 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1566, + "step": 26754 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 26755 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 26756 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 26757 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 26758 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 26759 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 26760 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1505, + "step": 26761 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1796, + "step": 26762 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 26763 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 26764 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 26765 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 26766 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 26767 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 26768 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 26769 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.0903, + "step": 26770 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 26771 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 26772 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.187, + "step": 26773 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1511, + "step": 26774 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 26775 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 26776 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 26777 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 26778 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 26779 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 26780 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.0917, + "step": 26781 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 26782 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 26783 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1557, + "step": 26784 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 26785 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 26786 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 26787 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 26788 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 26789 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 26790 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 26791 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 26792 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1465, + "step": 26793 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 26794 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 26795 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.0725, + "step": 26796 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 26797 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 26798 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 26799 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1666, + "step": 26800 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 26801 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1745, + "step": 26802 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 26803 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 26804 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1511, + "step": 26805 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 26806 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1485, + "step": 26807 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 26808 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.163, + "step": 26809 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 26810 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.176, + "step": 26811 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 26812 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 26813 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1643, + "step": 26814 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 26815 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 26816 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 26817 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 26818 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 26819 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1751, + "step": 26820 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1626, + "step": 26821 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 26822 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1655, + "step": 26823 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 26824 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 26825 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 26826 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 26827 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 26828 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 26829 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 26830 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 26831 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 26832 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 26833 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1535, + "step": 26834 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 26835 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 26836 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 26837 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 26838 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 26839 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1534, + "step": 26840 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1946, + "step": 26841 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 26842 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 26843 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1567, + "step": 26844 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1535, + "step": 26845 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 26846 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 26847 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 26848 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.0864, + "step": 26849 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1766, + "step": 26850 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 26851 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 26852 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 26853 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1456, + "step": 26854 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.0749, + "step": 26855 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 26856 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1453, + "step": 26857 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 26858 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 26859 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 26860 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1765, + "step": 26861 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 26862 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 26863 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 26864 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 26865 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 26866 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 26867 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 26868 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 26869 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 26870 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1758, + "step": 26871 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1583, + "step": 26872 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 26873 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 26874 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 26875 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1488, + "step": 26876 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 26877 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.184, + "step": 26878 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 26879 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 26880 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 26881 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 26882 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 26883 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 26884 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 26885 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 26886 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 26887 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1521, + "step": 26888 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 26889 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 26890 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 26891 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.0892, + "step": 26892 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 26893 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 26894 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 26895 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 26896 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 26897 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 26898 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 26899 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 26900 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 26901 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 26902 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1647, + "step": 26903 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 26904 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 26905 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 26906 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 26907 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 26908 + }, + { + "epoch": 1.63, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 26909 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 26910 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 26911 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 26912 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 26913 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1777, + "step": 26914 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 26915 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 26916 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 26917 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 26918 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 26919 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 26920 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1603, + "step": 26921 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1809, + "step": 26922 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 26923 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 26924 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 26925 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 26926 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1722, + "step": 26927 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1627, + "step": 26928 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.077, + "step": 26929 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 26930 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1577, + "step": 26931 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1657, + "step": 26932 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 26933 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 26934 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 26935 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 26936 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 26937 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 26938 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 26939 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 26940 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 26941 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 26942 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 26943 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 26944 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 26945 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 26946 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1511, + "step": 26947 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 26948 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 26949 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 26950 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1702, + "step": 26951 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 26952 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 26953 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1577, + "step": 26954 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 26955 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 26956 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 26957 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1459, + "step": 26958 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.174, + "step": 26959 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 26960 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 26961 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 26962 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 26963 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 26964 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1729, + "step": 26965 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1631, + "step": 26966 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.0828, + "step": 26967 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 26968 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 26969 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 26970 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 26971 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 26972 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 26973 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 26974 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 26975 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1836, + "step": 26976 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 26977 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 26978 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 26979 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 26980 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 26981 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 26982 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 26983 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 26984 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 26985 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1643, + "step": 26986 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1861, + "step": 26987 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 26988 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1732, + "step": 26989 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 26990 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.0883, + "step": 26991 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 26992 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 26993 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 26994 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 26995 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 26996 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1618, + "step": 26997 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 26998 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 26999 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 27000 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1633, + "step": 27001 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 27002 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 27003 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 27004 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 27005 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 27006 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 27007 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 27008 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 27009 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 27010 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1932, + "step": 27011 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1626, + "step": 27012 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 27013 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1751, + "step": 27014 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 27015 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 27016 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.153, + "step": 27017 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 27018 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.0901, + "step": 27019 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 27020 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1524, + "step": 27021 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 27022 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.2038, + "step": 27023 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 27024 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 27025 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 27026 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 27027 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 27028 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 27029 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 27030 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 27031 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 27032 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 27033 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 27034 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 27035 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 27036 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 27037 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 27038 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1657, + "step": 27039 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 27040 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 27041 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 27042 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 27043 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 27044 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 27045 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1611, + "step": 27046 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 27047 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 27048 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 27049 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 27050 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1794, + "step": 27051 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 27052 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 27053 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 27054 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1503, + "step": 27055 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1583, + "step": 27056 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 27057 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 27058 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 27059 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1733, + "step": 27060 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 27061 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1534, + "step": 27062 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1803, + "step": 27063 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1498, + "step": 27064 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 27065 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 27066 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 27067 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.084, + "step": 27068 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 27069 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 27070 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1633, + "step": 27071 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 27072 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 27073 + }, + { + "epoch": 1.64, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 27074 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1624, + "step": 27075 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 27076 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 27077 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 27078 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1515, + "step": 27079 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1561, + "step": 27080 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 27081 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 27082 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 27083 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 27084 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 27085 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 27086 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 27087 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 27088 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 27089 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 27090 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 27091 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 27092 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 27093 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 27094 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 27095 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1687, + "step": 27096 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 27097 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 27098 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 27099 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 27100 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 27101 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 27102 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.0811, + "step": 27103 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 27104 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 27105 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 27106 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 27107 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 27108 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1736, + "step": 27109 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 27110 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 27111 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 27112 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 27113 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 27114 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1589, + "step": 27115 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1631, + "step": 27116 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 27117 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 27118 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 27119 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1528, + "step": 27120 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 27121 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 27122 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 27123 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1616, + "step": 27124 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 27125 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 27126 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 27127 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1712, + "step": 27128 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 27129 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 27130 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 27131 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1562, + "step": 27132 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 27133 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1682, + "step": 27134 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 27135 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 27136 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1658, + "step": 27137 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1505, + "step": 27138 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 27139 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 27140 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 27141 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1566, + "step": 27142 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 27143 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 27144 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 27145 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 27146 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 27147 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 27148 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 27149 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 27150 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 27151 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1576, + "step": 27152 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 27153 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 27154 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.152, + "step": 27155 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 27156 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 27157 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.173, + "step": 27158 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 27159 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 27160 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 27161 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 27162 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 27163 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 27164 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1637, + "step": 27165 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1588, + "step": 27166 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 27167 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 27168 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 27169 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 27170 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 27171 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 27172 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 27173 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 27174 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 27175 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 27176 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1591, + "step": 27177 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 27178 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 27179 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1665, + "step": 27180 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 27181 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 27182 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 27183 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.148, + "step": 27184 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 27185 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 27186 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1637, + "step": 27187 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1532, + "step": 27188 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 27189 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 27190 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 27191 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 27192 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 27193 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1654, + "step": 27194 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 27195 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 27196 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 27197 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 27198 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 27199 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 27200 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 27201 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1714, + "step": 27202 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 27203 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 27204 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 27205 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 27206 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1719, + "step": 27207 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 27208 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.153, + "step": 27209 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 27210 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 27211 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 27212 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 27213 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 27214 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 27215 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 27216 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 27217 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 27218 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 27219 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 27220 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1637, + "step": 27221 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 27222 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1616, + "step": 27223 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 27224 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 27225 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 27226 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1873, + "step": 27227 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1801, + "step": 27228 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 27229 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 27230 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 27231 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 27232 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 27233 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 27234 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 27235 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 27236 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 27237 + }, + { + "epoch": 1.65, + "learning_rate": 2e-05, + "loss": 0.147, + "step": 27238 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 27239 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 27240 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 27241 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 27242 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 27243 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1643, + "step": 27244 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 27245 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1617, + "step": 27246 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 27247 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 27248 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 27249 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 27250 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 27251 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 27252 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 27253 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1626, + "step": 27254 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 27255 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.147, + "step": 27256 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 27257 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 27258 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 27259 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 27260 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 27261 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 27262 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 27263 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 27264 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 27265 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 27266 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1623, + "step": 27267 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 27268 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 27269 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 27270 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1554, + "step": 27271 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 27272 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 27273 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1604, + "step": 27274 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1752, + "step": 27275 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 27276 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1922, + "step": 27277 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.145, + "step": 27278 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1602, + "step": 27279 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 27280 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 27281 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 27282 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 27283 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1826, + "step": 27284 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 27285 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 27286 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 27287 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1606, + "step": 27288 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 27289 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1442, + "step": 27290 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 27291 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 27292 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 27293 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 27294 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 27295 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 27296 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 27297 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 27298 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 27299 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 27300 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1679, + "step": 27301 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 27302 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1746, + "step": 27303 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 27304 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.16, + "step": 27305 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 27306 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 27307 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 27308 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 27309 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 27310 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 27311 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 27312 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 27313 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1606, + "step": 27314 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 27315 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1611, + "step": 27316 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 27317 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1447, + "step": 27318 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 27319 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 27320 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 27321 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 27322 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 27323 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1596, + "step": 27324 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 27325 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1601, + "step": 27326 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 27327 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 27328 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 27329 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 27330 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 27331 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1528, + "step": 27332 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 27333 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 27334 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 27335 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1552, + "step": 27336 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 27337 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 27338 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 27339 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 27340 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1719, + "step": 27341 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1585, + "step": 27342 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1712, + "step": 27343 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 27344 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 27345 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 27346 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.0909, + "step": 27347 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 27348 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1538, + "step": 27349 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 27350 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 27351 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 27352 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 27353 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 27354 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 27355 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1688, + "step": 27356 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 27357 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 27358 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 27359 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1555, + "step": 27360 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 27361 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 27362 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 27363 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 27364 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 27365 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 27366 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1657, + "step": 27367 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 27368 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 27369 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 27370 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 27371 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 27372 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.155, + "step": 27373 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 27374 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1749, + "step": 27375 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1586, + "step": 27376 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 27377 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 27378 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 27379 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1699, + "step": 27380 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 27381 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 27382 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 27383 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 27384 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1443, + "step": 27385 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 27386 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 27387 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 27388 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 27389 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 27390 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 27391 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.199, + "step": 27392 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 27393 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 27394 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 27395 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1633, + "step": 27396 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1737, + "step": 27397 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 27398 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 27399 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 27400 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 27401 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 27402 + }, + { + "epoch": 1.66, + "learning_rate": 2e-05, + "loss": 0.146, + "step": 27403 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1733, + "step": 27404 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 27405 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 27406 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 27407 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1843, + "step": 27408 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 27409 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 27410 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 27411 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1652, + "step": 27412 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 27413 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 27414 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 27415 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1625, + "step": 27416 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 27417 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 27418 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 27419 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1601, + "step": 27420 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 27421 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 27422 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1676, + "step": 27423 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1485, + "step": 27424 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 27425 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 27426 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1495, + "step": 27427 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1725, + "step": 27428 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 27429 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1535, + "step": 27430 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 27431 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1586, + "step": 27432 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1585, + "step": 27433 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1524, + "step": 27434 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 27435 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 27436 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 27437 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 27438 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 27439 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 27440 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 27441 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 27442 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 27443 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1611, + "step": 27444 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 27445 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1495, + "step": 27446 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 27447 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 27448 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1731, + "step": 27449 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 27450 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 27451 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 27452 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 27453 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1695, + "step": 27454 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1705, + "step": 27455 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 27456 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 27457 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 27458 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 27459 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 27460 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 27461 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 27462 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 27463 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 27464 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 27465 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 27466 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.148, + "step": 27467 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1659, + "step": 27468 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 27469 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 27470 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 27471 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1633, + "step": 27472 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 27473 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 27474 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1621, + "step": 27475 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 27476 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 27477 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1783, + "step": 27478 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 27479 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 27480 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 27481 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 27482 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 27483 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 27484 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 27485 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 27486 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 27487 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 27488 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1688, + "step": 27489 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 27490 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 27491 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 27492 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.169, + "step": 27493 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 27494 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 27495 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 27496 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 27497 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 27498 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 27499 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 27500 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 27501 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 27502 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 27503 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 27504 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 27505 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 27506 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 27507 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 27508 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.0836, + "step": 27509 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 27510 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 27511 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 27512 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 27513 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 27514 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 27515 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 27516 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 27517 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 27518 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 27519 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 27520 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1621, + "step": 27521 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 27522 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 27523 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 27524 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 27525 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 27526 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1612, + "step": 27527 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 27528 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 27529 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 27530 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1659, + "step": 27531 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1602, + "step": 27532 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 27533 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1519, + "step": 27534 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 27535 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 27536 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1738, + "step": 27537 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 27538 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 27539 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 27540 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 27541 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 27542 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 27543 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 27544 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.2651, + "step": 27545 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 27546 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 27547 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 27548 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 27549 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 27550 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1814, + "step": 27551 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1798, + "step": 27552 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 27553 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 27554 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 27555 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 27556 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 27557 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 27558 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1429, + "step": 27559 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 27560 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 27561 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 27562 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1745, + "step": 27563 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.0936, + "step": 27564 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 27565 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 27566 + }, + { + "epoch": 1.67, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 27567 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 27568 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 27569 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1794, + "step": 27570 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 27571 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1614, + "step": 27572 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 27573 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 27574 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 27575 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 27576 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 27577 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1551, + "step": 27578 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 27579 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.184, + "step": 27580 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 27581 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 27582 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 27583 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 27584 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 27585 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.148, + "step": 27586 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1789, + "step": 27587 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 27588 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.2036, + "step": 27589 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.0899, + "step": 27590 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 27591 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1863, + "step": 27592 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 27593 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 27594 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 27595 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1687, + "step": 27596 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1468, + "step": 27597 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 27598 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 27599 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 27600 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 27601 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 27602 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 27603 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 27604 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 27605 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 27606 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 27607 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 27608 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 27609 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 27610 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.153, + "step": 27611 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 27612 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 27613 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 27614 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1618, + "step": 27615 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 27616 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 27617 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 27618 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1827, + "step": 27619 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 27620 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 27621 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1641, + "step": 27622 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1755, + "step": 27623 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 27624 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 27625 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.21, + "step": 27626 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 27627 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1557, + "step": 27628 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 27629 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 27630 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 27631 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1647, + "step": 27632 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 27633 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 27634 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 27635 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 27636 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 27637 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 27638 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 27639 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 27640 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 27641 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 27642 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 27643 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 27644 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1502, + "step": 27645 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 27646 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1804, + "step": 27647 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 27648 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 27649 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 27650 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 27651 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 27652 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 27653 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 27654 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 27655 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 27656 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 27657 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 27658 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 27659 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 27660 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 27661 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1566, + "step": 27662 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 27663 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1794, + "step": 27664 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 27665 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1992, + "step": 27666 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1576, + "step": 27667 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 27668 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 27669 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 27670 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 27671 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 27672 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1541, + "step": 27673 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 27674 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 27675 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 27676 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1635, + "step": 27677 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 27678 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 27679 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 27680 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 27681 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1525, + "step": 27682 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 27683 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 27684 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 27685 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 27686 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1635, + "step": 27687 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 27688 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 27689 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 27690 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 27691 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 27692 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 27693 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 27694 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1681, + "step": 27695 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 27696 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.156, + "step": 27697 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1714, + "step": 27698 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 27699 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 27700 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 27701 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 27702 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 27703 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 27704 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 27705 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1577, + "step": 27706 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 27707 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 27708 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1591, + "step": 27709 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 27710 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 27711 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1992, + "step": 27712 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1633, + "step": 27713 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 27714 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1679, + "step": 27715 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 27716 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 27717 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 27718 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1541, + "step": 27719 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 27720 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 27721 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 27722 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 27723 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 27724 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.144, + "step": 27725 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1859, + "step": 27726 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 27727 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 27728 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 27729 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 27730 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 27731 + }, + { + "epoch": 1.68, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 27732 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 27733 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 27734 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 27735 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1447, + "step": 27736 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.2, + "step": 27737 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 27738 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 27739 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1644, + "step": 27740 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 27741 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1497, + "step": 27742 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 27743 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.179, + "step": 27744 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 27745 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 27746 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 27747 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1723, + "step": 27748 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.148, + "step": 27749 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 27750 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 27751 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 27752 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 27753 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 27754 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 27755 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 27756 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1642, + "step": 27757 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 27758 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 27759 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 27760 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.0865, + "step": 27761 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 27762 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 27763 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 27764 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1555, + "step": 27765 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 27766 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1717, + "step": 27767 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1647, + "step": 27768 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 27769 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 27770 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 27771 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.0967, + "step": 27772 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 27773 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 27774 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1602, + "step": 27775 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 27776 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 27777 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 27778 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 27779 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 27780 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 27781 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 27782 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 27783 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 27784 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 27785 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 27786 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 27787 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 27788 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1787, + "step": 27789 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 27790 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 27791 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1762, + "step": 27792 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 27793 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 27794 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 27795 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 27796 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 27797 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1541, + "step": 27798 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1499, + "step": 27799 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1576, + "step": 27800 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 27801 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1958, + "step": 27802 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 27803 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1919, + "step": 27804 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1532, + "step": 27805 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 27806 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1855, + "step": 27807 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 27808 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 27809 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 27810 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 27811 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 27812 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 27813 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 27814 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1691, + "step": 27815 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 27816 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 27817 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 27818 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 27819 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 27820 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1707, + "step": 27821 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 27822 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 27823 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 27824 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 27825 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 27826 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 27827 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 27828 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 27829 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 27830 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 27831 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 27832 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 27833 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 27834 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 27835 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 27836 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 27837 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1538, + "step": 27838 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 27839 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1498, + "step": 27840 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 27841 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 27842 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1528, + "step": 27843 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 27844 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 27845 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 27846 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 27847 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 27848 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1453, + "step": 27849 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 27850 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1827, + "step": 27851 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 27852 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 27853 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 27854 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 27855 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 27856 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 27857 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1706, + "step": 27858 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 27859 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 27860 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 27861 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 27862 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 27863 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1566, + "step": 27864 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 27865 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 27866 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 27867 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 27868 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1579, + "step": 27869 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1468, + "step": 27870 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 27871 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1889, + "step": 27872 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 27873 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 27874 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 27875 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 27876 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 27877 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 27878 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 27879 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 27880 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 27881 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.156, + "step": 27882 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 27883 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1868, + "step": 27884 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 27885 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1974, + "step": 27886 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 27887 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 27888 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 27889 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1527, + "step": 27890 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 27891 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 27892 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 27893 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 27894 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 27895 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 27896 + }, + { + "epoch": 1.69, + "learning_rate": 2e-05, + "loss": 0.1525, + "step": 27897 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 27898 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 27899 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 27900 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 27901 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.2253, + "step": 27902 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 27903 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 27904 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 27905 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 27906 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1638, + "step": 27907 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 27908 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.165, + "step": 27909 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 27910 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1561, + "step": 27911 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1998, + "step": 27912 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 27913 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 27914 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 27915 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 27916 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1635, + "step": 27917 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 27918 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 27919 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1765, + "step": 27920 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 27921 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1532, + "step": 27922 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 27923 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1505, + "step": 27924 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 27925 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 27926 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 27927 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 27928 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 27929 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 27930 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 27931 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 27932 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1495, + "step": 27933 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 27934 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 27935 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1597, + "step": 27936 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 27937 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 27938 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1516, + "step": 27939 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 27940 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1806, + "step": 27941 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1518, + "step": 27942 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 27943 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 27944 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 27945 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1876, + "step": 27946 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1947, + "step": 27947 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 27948 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 27949 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 27950 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 27951 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 27952 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 27953 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1618, + "step": 27954 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 27955 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1661, + "step": 27956 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 27957 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.0864, + "step": 27958 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.0952, + "step": 27959 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 27960 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 27961 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 27962 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 27963 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 27964 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 27965 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 27966 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 27967 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.0911, + "step": 27968 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 27969 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 27970 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1524, + "step": 27971 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1586, + "step": 27972 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 27973 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 27974 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1878, + "step": 27975 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 27976 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 27977 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 27978 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1598, + "step": 27979 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 27980 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 27981 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 27982 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 27983 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1518, + "step": 27984 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 27985 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 27986 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 27987 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 27988 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1812, + "step": 27989 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1588, + "step": 27990 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1502, + "step": 27991 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1531, + "step": 27992 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 27993 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 27994 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 27995 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 27996 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1519, + "step": 27997 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 27998 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 27999 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.175, + "step": 28000 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 28001 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 28002 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 28003 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 28004 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 28005 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 28006 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1527, + "step": 28007 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 28008 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 28009 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 28010 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 28011 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1676, + "step": 28012 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1923, + "step": 28013 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 28014 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 28015 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 28016 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 28017 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 28018 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 28019 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 28020 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 28021 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 28022 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 28023 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 28024 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 28025 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1703, + "step": 28026 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 28027 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 28028 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 28029 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 28030 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 28031 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 28032 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 28033 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 28034 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1515, + "step": 28035 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.0967, + "step": 28036 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 28037 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 28038 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 28039 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 28040 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 28041 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.159, + "step": 28042 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 28043 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1526, + "step": 28044 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 28045 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 28046 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 28047 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 28048 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 28049 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.0782, + "step": 28050 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 28051 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 28052 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 28053 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 28054 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 28055 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 28056 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 28057 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1519, + "step": 28058 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 28059 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 28060 + }, + { + "epoch": 1.7, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 28061 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 28062 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 28063 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 28064 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 28065 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 28066 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 28067 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 28068 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 28069 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 28070 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 28071 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 28072 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 28073 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1532, + "step": 28074 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1468, + "step": 28075 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1533, + "step": 28076 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 28077 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 28078 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1546, + "step": 28079 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1645, + "step": 28080 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 28081 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 28082 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 28083 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1443, + "step": 28084 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 28085 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.162, + "step": 28086 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 28087 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 28088 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1465, + "step": 28089 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 28090 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 28091 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1746, + "step": 28092 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 28093 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 28094 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 28095 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1734, + "step": 28096 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 28097 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.155, + "step": 28098 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 28099 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 28100 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1876, + "step": 28101 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 28102 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 28103 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 28104 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 28105 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 28106 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 28107 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 28108 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1478, + "step": 28109 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 28110 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 28111 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 28112 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 28113 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 28114 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 28115 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1521, + "step": 28116 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1724, + "step": 28117 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 28118 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 28119 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 28120 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 28121 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 28122 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 28123 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 28124 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 28125 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 28126 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 28127 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 28128 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 28129 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1561, + "step": 28130 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 28131 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 28132 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 28133 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 28134 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 28135 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1574, + "step": 28136 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 28137 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 28138 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 28139 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 28140 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 28141 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 28142 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1757, + "step": 28143 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1462, + "step": 28144 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1442, + "step": 28145 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1773, + "step": 28146 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1521, + "step": 28147 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 28148 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 28149 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 28150 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 28151 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 28152 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1686, + "step": 28153 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 28154 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 28155 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1963, + "step": 28156 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 28157 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 28158 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 28159 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 28160 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 28161 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 28162 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 28163 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1926, + "step": 28164 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 28165 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 28166 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1429, + "step": 28167 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 28168 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 28169 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 28170 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 28171 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 28172 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 28173 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 28174 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 28175 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1516, + "step": 28176 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 28177 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 28178 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1632, + "step": 28179 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 28180 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1608, + "step": 28181 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 28182 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1488, + "step": 28183 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 28184 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 28185 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 28186 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 28187 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 28188 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 28189 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 28190 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 28191 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 28192 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 28193 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 28194 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 28195 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 28196 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 28197 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 28198 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 28199 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 28200 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 28201 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 28202 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 28203 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.0903, + "step": 28204 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 28205 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 28206 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 28207 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 28208 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 28209 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 28210 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1429, + "step": 28211 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 28212 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 28213 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 28214 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 28215 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 28216 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 28217 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 28218 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1797, + "step": 28219 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 28220 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1617, + "step": 28221 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 28222 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 28223 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1622, + "step": 28224 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 28225 + }, + { + "epoch": 1.71, + "learning_rate": 2e-05, + "loss": 0.1582, + "step": 28226 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 28227 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 28228 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 28229 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 28230 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 28231 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 28232 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1714, + "step": 28233 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 28234 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 28235 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 28236 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1524, + "step": 28237 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 28238 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.0838, + "step": 28239 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 28240 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1671, + "step": 28241 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 28242 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 28243 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 28244 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 28245 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 28246 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1516, + "step": 28247 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 28248 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 28249 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1447, + "step": 28250 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 28251 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.148, + "step": 28252 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 28253 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 28254 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 28255 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1829, + "step": 28256 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 28257 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 28258 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 28259 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1629, + "step": 28260 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 28261 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 28262 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 28263 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 28264 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 28265 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.0941, + "step": 28266 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 28267 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1507, + "step": 28268 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 28269 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1478, + "step": 28270 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 28271 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 28272 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 28273 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 28274 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 28275 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 28276 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 28277 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 28278 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 28279 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 28280 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 28281 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 28282 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 28283 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 28284 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 28285 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 28286 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.164, + "step": 28287 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 28288 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 28289 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 28290 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 28291 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 28292 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 28293 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 28294 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 28295 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 28296 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 28297 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 28298 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 28299 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 28300 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 28301 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 28302 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1714, + "step": 28303 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 28304 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 28305 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 28306 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1535, + "step": 28307 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 28308 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1469, + "step": 28309 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 28310 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1456, + "step": 28311 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 28312 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 28313 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 28314 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 28315 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 28316 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 28317 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 28318 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 28319 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 28320 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 28321 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.152, + "step": 28322 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.159, + "step": 28323 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 28324 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1583, + "step": 28325 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 28326 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 28327 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 28328 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 28329 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 28330 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1603, + "step": 28331 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 28332 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 28333 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1821, + "step": 28334 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 28335 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 28336 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 28337 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1654, + "step": 28338 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 28339 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 28340 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1748, + "step": 28341 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 28342 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 28343 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 28344 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 28345 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 28346 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1586, + "step": 28347 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 28348 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 28349 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 28350 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 28351 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1476, + "step": 28352 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 28353 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1592, + "step": 28354 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 28355 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1699, + "step": 28356 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1753, + "step": 28357 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 28358 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 28359 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 28360 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 28361 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 28362 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 28363 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1456, + "step": 28364 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 28365 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.0864, + "step": 28366 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 28367 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 28368 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 28369 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 28370 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 28371 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1622, + "step": 28372 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 28373 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 28374 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 28375 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 28376 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 28377 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 28378 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 28379 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 28380 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 28381 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1666, + "step": 28382 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1726, + "step": 28383 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 28384 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 28385 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 28386 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 28387 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 28388 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 28389 + }, + { + "epoch": 1.72, + "learning_rate": 2e-05, + "loss": 0.1585, + "step": 28390 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 28391 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 28392 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 28393 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 28394 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 28395 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 28396 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 28397 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1889, + "step": 28398 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1498, + "step": 28399 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 28400 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 28401 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 28402 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1661, + "step": 28403 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 28404 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1704, + "step": 28405 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 28406 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 28407 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 28408 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 28409 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 28410 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1742, + "step": 28411 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 28412 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 28413 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1586, + "step": 28414 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 28415 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 28416 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 28417 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 28418 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1827, + "step": 28419 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 28420 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1485, + "step": 28421 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 28422 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 28423 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 28424 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.152, + "step": 28425 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 28426 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 28427 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1462, + "step": 28428 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 28429 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 28430 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 28431 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 28432 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 28433 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 28434 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1476, + "step": 28435 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 28436 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 28437 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 28438 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 28439 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 28440 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 28441 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 28442 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 28443 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 28444 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1517, + "step": 28445 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 28446 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 28447 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.207, + "step": 28448 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1478, + "step": 28449 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1583, + "step": 28450 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 28451 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1576, + "step": 28452 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 28453 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 28454 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 28455 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 28456 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 28457 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 28458 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1774, + "step": 28459 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 28460 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 28461 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 28462 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 28463 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 28464 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 28465 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 28466 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.0967, + "step": 28467 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1627, + "step": 28468 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.167, + "step": 28469 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 28470 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.171, + "step": 28471 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 28472 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 28473 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 28474 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 28475 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 28476 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 28477 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 28478 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 28479 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1854, + "step": 28480 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 28481 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 28482 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 28483 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 28484 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.185, + "step": 28485 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1533, + "step": 28486 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 28487 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 28488 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 28489 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 28490 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 28491 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.153, + "step": 28492 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 28493 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.169, + "step": 28494 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 28495 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 28496 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1871, + "step": 28497 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 28498 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 28499 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1563, + "step": 28500 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1942, + "step": 28501 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1621, + "step": 28502 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 28503 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 28504 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.0895, + "step": 28505 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1578, + "step": 28506 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 28507 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 28508 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 28509 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 28510 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 28511 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 28512 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 28513 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 28514 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1662, + "step": 28515 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 28516 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1596, + "step": 28517 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 28518 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1801, + "step": 28519 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 28520 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 28521 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 28522 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1623, + "step": 28523 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 28524 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 28525 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 28526 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 28527 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1502, + "step": 28528 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1552, + "step": 28529 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 28530 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 28531 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.0834, + "step": 28532 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1586, + "step": 28533 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1505, + "step": 28534 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 28535 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 28536 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1929, + "step": 28537 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 28538 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1774, + "step": 28539 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 28540 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 28541 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1532, + "step": 28542 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 28543 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 28544 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1881, + "step": 28545 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 28546 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 28547 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 28548 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 28549 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 28550 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1534, + "step": 28551 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 28552 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 28553 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1708, + "step": 28554 + }, + { + "epoch": 1.73, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 28555 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 28556 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 28557 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 28558 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 28559 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 28560 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 28561 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 28562 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 28563 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 28564 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 28565 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1723, + "step": 28566 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 28567 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 28568 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 28569 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1845, + "step": 28570 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 28571 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 28572 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 28573 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 28574 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 28575 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 28576 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 28577 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 28578 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 28579 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 28580 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 28581 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 28582 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 28583 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 28584 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1443, + "step": 28585 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1558, + "step": 28586 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.2055, + "step": 28587 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 28588 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 28589 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 28590 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1476, + "step": 28591 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1511, + "step": 28592 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 28593 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 28594 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 28595 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 28596 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 28597 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 28598 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 28599 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 28600 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1551, + "step": 28601 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 28602 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 28603 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 28604 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 28605 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 28606 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1549, + "step": 28607 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 28608 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 28609 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 28610 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 28611 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 28612 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 28613 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 28614 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 28615 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 28616 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1621, + "step": 28617 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 28618 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 28619 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 28620 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 28621 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 28622 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1611, + "step": 28623 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 28624 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 28625 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 28626 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 28627 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.0887, + "step": 28628 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 28629 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 28630 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 28631 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 28632 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1538, + "step": 28633 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 28634 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1745, + "step": 28635 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 28636 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 28637 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 28638 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 28639 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 28640 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 28641 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1642, + "step": 28642 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 28643 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 28644 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 28645 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.171, + "step": 28646 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 28647 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 28648 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 28649 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1633, + "step": 28650 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 28651 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 28652 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 28653 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 28654 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 28655 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1507, + "step": 28656 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 28657 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1705, + "step": 28658 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1533, + "step": 28659 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 28660 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 28661 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 28662 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.144, + "step": 28663 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 28664 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 28665 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 28666 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 28667 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 28668 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 28669 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1729, + "step": 28670 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 28671 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 28672 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 28673 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 28674 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1852, + "step": 28675 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 28676 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 28677 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1773, + "step": 28678 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 28679 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 28680 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 28681 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 28682 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.2023, + "step": 28683 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 28684 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 28685 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 28686 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 28687 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 28688 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 28689 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 28690 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 28691 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 28692 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1741, + "step": 28693 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 28694 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 28695 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 28696 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 28697 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 28698 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 28699 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1612, + "step": 28700 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 28701 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 28702 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 28703 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 28704 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 28705 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 28706 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 28707 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 28708 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 28709 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 28710 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1515, + "step": 28711 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 28712 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.167, + "step": 28713 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 28714 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1542, + "step": 28715 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1511, + "step": 28716 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 28717 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 28718 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 28719 + }, + { + "epoch": 1.74, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 28720 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 28721 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 28722 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 28723 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1897, + "step": 28724 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 28725 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.0807, + "step": 28726 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 28727 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1753, + "step": 28728 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 28729 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 28730 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 28731 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 28732 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 28733 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 28734 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 28735 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 28736 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 28737 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 28738 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1518, + "step": 28739 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 28740 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 28741 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 28742 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 28743 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 28744 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 28745 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1661, + "step": 28746 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 28747 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 28748 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 28749 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 28750 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 28751 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 28752 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1645, + "step": 28753 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1612, + "step": 28754 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 28755 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 28756 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1747, + "step": 28757 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 28758 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 28759 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1635, + "step": 28760 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 28761 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 28762 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 28763 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 28764 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 28765 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1533, + "step": 28766 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 28767 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1503, + "step": 28768 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 28769 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 28770 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 28771 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1694, + "step": 28772 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 28773 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 28774 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 28775 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 28776 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1741, + "step": 28777 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 28778 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 28779 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 28780 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 28781 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 28782 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1577, + "step": 28783 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1561, + "step": 28784 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 28785 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 28786 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 28787 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 28788 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 28789 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 28790 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 28791 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 28792 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.0886, + "step": 28793 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 28794 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 28795 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1667, + "step": 28796 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 28797 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 28798 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1643, + "step": 28799 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 28800 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 28801 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 28802 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1714, + "step": 28803 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 28804 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 28805 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 28806 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 28807 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 28808 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 28809 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 28810 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 28811 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.088, + "step": 28812 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1805, + "step": 28813 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 28814 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 28815 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 28816 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 28817 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 28818 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 28819 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 28820 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 28821 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 28822 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 28823 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 28824 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 28825 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 28826 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 28827 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 28828 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1515, + "step": 28829 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 28830 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.147, + "step": 28831 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 28832 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 28833 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 28834 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1469, + "step": 28835 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 28836 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 28837 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 28838 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 28839 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 28840 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 28841 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 28842 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1497, + "step": 28843 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 28844 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 28845 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 28846 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 28847 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 28848 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 28849 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1519, + "step": 28850 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 28851 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 28852 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 28853 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 28854 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 28855 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 28856 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.167, + "step": 28857 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 28858 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 28859 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1519, + "step": 28860 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 28861 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 28862 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1805, + "step": 28863 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 28864 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 28865 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 28866 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.169, + "step": 28867 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 28868 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 28869 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1764, + "step": 28870 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 28871 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 28872 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 28873 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1564, + "step": 28874 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 28875 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 28876 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 28877 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 28878 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 28879 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 28880 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1762, + "step": 28881 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 28882 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 28883 + }, + { + "epoch": 1.75, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 28884 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1586, + "step": 28885 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 28886 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 28887 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 28888 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 28889 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1542, + "step": 28890 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 28891 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1533, + "step": 28892 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 28893 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 28894 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 28895 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 28896 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 28897 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 28898 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 28899 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 28900 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 28901 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 28902 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 28903 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.144, + "step": 28904 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1726, + "step": 28905 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 28906 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 28907 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 28908 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.161, + "step": 28909 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 28910 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 28911 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1942, + "step": 28912 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 28913 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 28914 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 28915 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1792, + "step": 28916 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 28917 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 28918 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1669, + "step": 28919 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 28920 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1633, + "step": 28921 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 28922 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 28923 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.0911, + "step": 28924 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 28925 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 28926 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 28927 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 28928 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 28929 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 28930 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1591, + "step": 28931 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 28932 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 28933 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 28934 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 28935 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 28936 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 28937 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 28938 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 28939 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.0917, + "step": 28940 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 28941 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1811, + "step": 28942 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 28943 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 28944 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 28945 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 28946 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1546, + "step": 28947 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 28948 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 28949 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1538, + "step": 28950 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 28951 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 28952 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 28953 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 28954 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 28955 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 28956 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 28957 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 28958 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 28959 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.144, + "step": 28960 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1829, + "step": 28961 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 28962 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 28963 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1568, + "step": 28964 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 28965 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1731, + "step": 28966 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 28967 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 28968 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 28969 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 28970 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 28971 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 28972 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1469, + "step": 28973 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1697, + "step": 28974 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1603, + "step": 28975 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 28976 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 28977 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 28978 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 28979 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 28980 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 28981 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 28982 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 28983 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 28984 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 28985 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 28986 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 28987 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 28988 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 28989 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 28990 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 28991 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 28992 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 28993 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 28994 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 28995 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 28996 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 28997 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1823, + "step": 28998 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1685, + "step": 28999 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 29000 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 29001 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 29002 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 29003 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 29004 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1729, + "step": 29005 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 29006 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 29007 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 29008 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 29009 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 29010 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 29011 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1659, + "step": 29012 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1614, + "step": 29013 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 29014 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 29015 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 29016 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1507, + "step": 29017 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 29018 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1629, + "step": 29019 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 29020 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 29021 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 29022 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 29023 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 29024 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 29025 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.146, + "step": 29026 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 29027 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 29028 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 29029 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1469, + "step": 29030 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1757, + "step": 29031 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 29032 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 29033 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 29034 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 29035 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 29036 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 29037 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 29038 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 29039 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1773, + "step": 29040 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1604, + "step": 29041 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 29042 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 29043 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 29044 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 29045 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1948, + "step": 29046 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 29047 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 29048 + }, + { + "epoch": 1.76, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 29049 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.2005, + "step": 29050 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 29051 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 29052 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 29053 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 29054 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 29055 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 29056 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 29057 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.153, + "step": 29058 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 29059 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.182, + "step": 29060 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1624, + "step": 29061 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 29062 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 29063 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 29064 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 29065 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 29066 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 29067 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 29068 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 29069 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1674, + "step": 29070 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 29071 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 29072 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 29073 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1651, + "step": 29074 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 29075 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 29076 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1656, + "step": 29077 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 29078 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1538, + "step": 29079 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 29080 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 29081 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 29082 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 29083 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 29084 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1549, + "step": 29085 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 29086 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 29087 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 29088 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 29089 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 29090 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1768, + "step": 29091 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1605, + "step": 29092 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1528, + "step": 29093 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1718, + "step": 29094 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 29095 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.16, + "step": 29096 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 29097 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 29098 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.0863, + "step": 29099 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 29100 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 29101 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 29102 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 29103 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 29104 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 29105 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 29106 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1456, + "step": 29107 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1776, + "step": 29108 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 29109 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 29110 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 29111 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 29112 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 29113 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1502, + "step": 29114 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 29115 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1676, + "step": 29116 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 29117 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 29118 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1606, + "step": 29119 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.153, + "step": 29120 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.162, + "step": 29121 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1563, + "step": 29122 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1627, + "step": 29123 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 29124 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 29125 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 29126 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 29127 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 29128 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 29129 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 29130 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 29131 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1502, + "step": 29132 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 29133 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 29134 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.0932, + "step": 29135 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 29136 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 29137 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1624, + "step": 29138 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 29139 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 29140 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 29141 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 29142 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 29143 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 29144 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 29145 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 29146 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 29147 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 29148 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 29149 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 29150 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 29151 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 29152 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 29153 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 29154 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 29155 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 29156 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 29157 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 29158 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1815, + "step": 29159 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 29160 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 29161 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 29162 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 29163 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1561, + "step": 29164 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 29165 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1604, + "step": 29166 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 29167 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 29168 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1612, + "step": 29169 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 29170 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 29171 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1607, + "step": 29172 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1775, + "step": 29173 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.188, + "step": 29174 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 29175 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 29176 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 29177 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 29178 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1495, + "step": 29179 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 29180 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1634, + "step": 29181 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 29182 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 29183 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 29184 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 29185 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1429, + "step": 29186 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 29187 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 29188 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.145, + "step": 29189 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 29190 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 29191 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 29192 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 29193 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 29194 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1672, + "step": 29195 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1558, + "step": 29196 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 29197 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 29198 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 29199 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 29200 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.161, + "step": 29201 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 29202 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 29203 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 29204 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 29205 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 29206 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 29207 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 29208 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1596, + "step": 29209 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1635, + "step": 29210 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 29211 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 29212 + }, + { + "epoch": 1.77, + "learning_rate": 2e-05, + "loss": 0.1596, + "step": 29213 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 29214 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 29215 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 29216 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1704, + "step": 29217 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1646, + "step": 29218 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 29219 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 29220 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 29221 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.153, + "step": 29222 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1513, + "step": 29223 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 29224 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 29225 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1721, + "step": 29226 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1429, + "step": 29227 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 29228 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 29229 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 29230 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1687, + "step": 29231 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 29232 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 29233 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.169, + "step": 29234 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 29235 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 29236 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 29237 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 29238 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 29239 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 29240 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 29241 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 29242 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 29243 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 29244 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 29245 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.2116, + "step": 29246 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1451, + "step": 29247 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 29248 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 29249 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1828, + "step": 29250 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 29251 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 29252 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 29253 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 29254 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 29255 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 29256 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 29257 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 29258 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 29259 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 29260 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 29261 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 29262 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 29263 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 29264 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 29265 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 29266 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 29267 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1468, + "step": 29268 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 29269 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 29270 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 29271 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 29272 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 29273 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 29274 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 29275 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 29276 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 29277 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 29278 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 29279 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 29280 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1857, + "step": 29281 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 29282 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 29283 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 29284 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 29285 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 29286 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 29287 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 29288 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1513, + "step": 29289 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1456, + "step": 29290 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 29291 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1867, + "step": 29292 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 29293 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.158, + "step": 29294 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 29295 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1794, + "step": 29296 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 29297 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1532, + "step": 29298 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 29299 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1518, + "step": 29300 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 29301 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 29302 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 29303 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1692, + "step": 29304 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 29305 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 29306 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1562, + "step": 29307 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1785, + "step": 29308 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 29309 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 29310 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 29311 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 29312 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 29313 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 29314 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 29315 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 29316 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 29317 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 29318 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 29319 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 29320 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 29321 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.2055, + "step": 29322 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 29323 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 29324 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 29325 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 29326 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 29327 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 29328 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1606, + "step": 29329 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.0906, + "step": 29330 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 29331 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 29332 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 29333 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 29334 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 29335 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 29336 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 29337 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 29338 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 29339 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1718, + "step": 29340 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 29341 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 29342 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 29343 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 29344 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 29345 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 29346 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1707, + "step": 29347 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 29348 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 29349 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 29350 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 29351 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 29352 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 29353 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 29354 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 29355 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 29356 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 29357 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 29358 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 29359 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 29360 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 29361 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1526, + "step": 29362 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1716, + "step": 29363 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 29364 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1552, + "step": 29365 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1772, + "step": 29366 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 29367 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 29368 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 29369 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 29370 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 29371 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 29372 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 29373 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 29374 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 29375 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1469, + "step": 29376 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 29377 + }, + { + "epoch": 1.78, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 29378 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1682, + "step": 29379 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 29380 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 29381 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 29382 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 29383 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 29384 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 29385 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 29386 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 29387 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 29388 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 29389 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.0799, + "step": 29390 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1593, + "step": 29391 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 29392 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 29393 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 29394 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1498, + "step": 29395 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 29396 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 29397 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 29398 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 29399 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1804, + "step": 29400 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 29401 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 29402 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 29403 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1451, + "step": 29404 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 29405 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1634, + "step": 29406 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 29407 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1612, + "step": 29408 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 29409 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 29410 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 29411 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 29412 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1465, + "step": 29413 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 29414 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 29415 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 29416 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1811, + "step": 29417 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 29418 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 29419 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 29420 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 29421 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 29422 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 29423 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 29424 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 29425 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.168, + "step": 29426 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 29427 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1549, + "step": 29428 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 29429 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 29430 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 29431 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 29432 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 29433 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1599, + "step": 29434 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1578, + "step": 29435 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 29436 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 29437 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1497, + "step": 29438 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 29439 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 29440 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 29441 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1528, + "step": 29442 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 29443 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1725, + "step": 29444 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 29445 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1798, + "step": 29446 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 29447 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1649, + "step": 29448 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.169, + "step": 29449 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1465, + "step": 29450 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 29451 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 29452 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 29453 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1532, + "step": 29454 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 29455 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 29456 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 29457 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 29458 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1829, + "step": 29459 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 29460 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.2, + "step": 29461 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1618, + "step": 29462 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 29463 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 29464 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 29465 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.147, + "step": 29466 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 29467 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1683, + "step": 29468 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 29469 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 29470 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 29471 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 29472 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 29473 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 29474 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 29475 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 29476 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 29477 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1456, + "step": 29478 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 29479 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 29480 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 29481 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 29482 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1714, + "step": 29483 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 29484 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 29485 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1597, + "step": 29486 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 29487 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 29488 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 29489 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 29490 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 29491 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 29492 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 29493 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 29494 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 29495 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 29496 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 29497 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 29498 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 29499 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 29500 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.181, + "step": 29501 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 29502 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 29503 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 29504 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1507, + "step": 29505 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1595, + "step": 29506 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 29507 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1603, + "step": 29508 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 29509 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 29510 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 29511 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1644, + "step": 29512 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.168, + "step": 29513 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 29514 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 29515 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 29516 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 29517 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 29518 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 29519 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 29520 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 29521 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1456, + "step": 29522 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 29523 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 29524 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 29525 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1638, + "step": 29526 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 29527 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1611, + "step": 29528 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.169, + "step": 29529 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 29530 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 29531 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 29532 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 29533 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 29534 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 29535 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 29536 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 29537 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 29538 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 29539 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1866, + "step": 29540 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 29541 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1531, + "step": 29542 + }, + { + "epoch": 1.79, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 29543 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 29544 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 29545 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 29546 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 29547 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 29548 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.0872, + "step": 29549 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 29550 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 29551 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 29552 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 29553 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 29554 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 29555 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 29556 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.167, + "step": 29557 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 29558 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1498, + "step": 29559 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 29560 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 29561 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 29562 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.164, + "step": 29563 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 29564 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.158, + "step": 29565 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 29566 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 29567 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 29568 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 29569 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 29570 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 29571 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 29572 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 29573 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 29574 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 29575 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 29576 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.161, + "step": 29577 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1622, + "step": 29578 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 29579 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 29580 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 29581 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.0943, + "step": 29582 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 29583 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 29584 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 29585 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 29586 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 29587 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 29588 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1453, + "step": 29589 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.0889, + "step": 29590 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1848, + "step": 29591 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 29592 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 29593 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 29594 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 29595 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 29596 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1549, + "step": 29597 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1694, + "step": 29598 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 29599 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 29600 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 29601 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 29602 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 29603 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 29604 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1468, + "step": 29605 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 29606 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1997, + "step": 29607 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 29608 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1498, + "step": 29609 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.0935, + "step": 29610 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1802, + "step": 29611 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 29612 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.0855, + "step": 29613 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1651, + "step": 29614 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 29615 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 29616 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 29617 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 29618 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 29619 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 29620 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 29621 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 29622 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1654, + "step": 29623 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 29624 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1803, + "step": 29625 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 29626 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 29627 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 29628 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 29629 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 29630 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 29631 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 29632 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 29633 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 29634 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 29635 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 29636 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 29637 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 29638 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1515, + "step": 29639 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 29640 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 29641 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 29642 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 29643 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 29644 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.0869, + "step": 29645 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1485, + "step": 29646 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 29647 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 29648 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 29649 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 29650 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 29651 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 29652 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 29653 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1542, + "step": 29654 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 29655 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 29656 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 29657 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 29658 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 29659 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 29660 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 29661 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 29662 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 29663 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 29664 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 29665 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1498, + "step": 29666 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1488, + "step": 29667 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 29668 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 29669 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 29670 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 29671 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 29672 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 29673 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 29674 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 29675 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 29676 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 29677 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 29678 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 29679 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 29680 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 29681 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1497, + "step": 29682 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 29683 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 29684 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 29685 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 29686 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 29687 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.148, + "step": 29688 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 29689 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 29690 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 29691 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 29692 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 29693 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 29694 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 29695 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 29696 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 29697 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 29698 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1675, + "step": 29699 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1516, + "step": 29700 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 29701 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 29702 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 29703 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 29704 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 29705 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 29706 + }, + { + "epoch": 1.8, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 29707 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 29708 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 29709 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 29710 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 29711 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 29712 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 29713 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 29714 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 29715 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 29716 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.144, + "step": 29717 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 29718 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 29719 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 29720 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 29721 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 29722 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 29723 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1578, + "step": 29724 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1604, + "step": 29725 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 29726 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1616, + "step": 29727 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 29728 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 29729 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 29730 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 29731 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 29732 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 29733 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 29734 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 29735 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 29736 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 29737 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 29738 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1778, + "step": 29739 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 29740 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 29741 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1566, + "step": 29742 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 29743 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 29744 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 29745 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 29746 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 29747 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 29748 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 29749 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 29750 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.144, + "step": 29751 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 29752 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 29753 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 29754 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 29755 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 29756 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 29757 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 29758 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 29759 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 29760 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 29761 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1745, + "step": 29762 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 29763 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 29764 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 29765 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.204, + "step": 29766 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.155, + "step": 29767 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 29768 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 29769 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 29770 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 29771 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 29772 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 29773 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 29774 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 29775 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 29776 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 29777 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 29778 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 29779 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 29780 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 29781 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 29782 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 29783 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 29784 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 29785 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 29786 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 29787 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 29788 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1695, + "step": 29789 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 29790 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 29791 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.0824, + "step": 29792 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 29793 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 29794 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 29795 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 29796 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 29797 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 29798 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1613, + "step": 29799 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1542, + "step": 29800 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 29801 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 29802 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 29803 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 29804 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 29805 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 29806 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 29807 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 29808 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 29809 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 29810 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.2272, + "step": 29811 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 29812 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1574, + "step": 29813 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 29814 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 29815 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 29816 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 29817 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1783, + "step": 29818 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 29819 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 29820 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1664, + "step": 29821 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 29822 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 29823 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 29824 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 29825 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 29826 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 29827 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1613, + "step": 29828 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 29829 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 29830 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 29831 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 29832 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 29833 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 29834 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 29835 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 29836 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.195, + "step": 29837 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 29838 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 29839 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 29840 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 29841 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 29842 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 29843 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1442, + "step": 29844 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 29845 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 29846 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 29847 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 29848 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 29849 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 29850 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 29851 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 29852 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 29853 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 29854 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1485, + "step": 29855 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 29856 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 29857 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1566, + "step": 29858 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 29859 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 29860 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 29861 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 29862 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 29863 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 29864 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 29865 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 29866 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 29867 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 29868 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 29869 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 29870 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 29871 + }, + { + "epoch": 1.81, + "learning_rate": 2e-05, + "loss": 0.182, + "step": 29872 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1705, + "step": 29873 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 29874 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 29875 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 29876 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 29877 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 29878 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 29879 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1576, + "step": 29880 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 29881 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 29882 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 29883 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 29884 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 29885 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 29886 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 29887 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.145, + "step": 29888 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1511, + "step": 29889 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1756, + "step": 29890 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 29891 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 29892 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 29893 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 29894 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 29895 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 29896 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 29897 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1664, + "step": 29898 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 29899 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1605, + "step": 29900 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.0936, + "step": 29901 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 29902 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 29903 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 29904 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 29905 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 29906 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 29907 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 29908 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 29909 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 29910 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 29911 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 29912 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 29913 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 29914 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1879, + "step": 29915 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1563, + "step": 29916 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 29917 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 29918 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 29919 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 29920 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 29921 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1567, + "step": 29922 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 29923 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 29924 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 29925 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 29926 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.158, + "step": 29927 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1668, + "step": 29928 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 29929 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 29930 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 29931 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 29932 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 29933 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1596, + "step": 29934 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 29935 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1551, + "step": 29936 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 29937 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 29938 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 29939 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 29940 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 29941 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 29942 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 29943 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.159, + "step": 29944 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 29945 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 29946 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1878, + "step": 29947 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 29948 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 29949 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1692, + "step": 29950 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 29951 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 29952 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 29953 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 29954 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1843, + "step": 29955 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 29956 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 29957 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1761, + "step": 29958 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 29959 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 29960 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 29961 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 29962 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1551, + "step": 29963 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1607, + "step": 29964 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 29965 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 29966 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 29967 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 29968 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1671, + "step": 29969 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 29970 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.146, + "step": 29971 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 29972 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 29973 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 29974 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 29975 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 29976 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 29977 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1604, + "step": 29978 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 29979 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 29980 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 29981 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 29982 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 29983 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 29984 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 29985 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 29986 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 29987 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 29988 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1652, + "step": 29989 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1668, + "step": 29990 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 29991 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 29992 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 29993 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 29994 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 29995 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 29996 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1586, + "step": 29997 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 29998 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 29999 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 30000 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 30001 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1447, + "step": 30002 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 30003 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 30004 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 30005 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 30006 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 30007 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 30008 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1546, + "step": 30009 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 30010 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 30011 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 30012 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 30013 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 30014 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 30015 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 30016 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 30017 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 30018 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 30019 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 30020 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 30021 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 30022 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1623, + "step": 30023 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1726, + "step": 30024 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 30025 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 30026 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 30027 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1891, + "step": 30028 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 30029 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1979, + "step": 30030 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 30031 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 30032 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 30033 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 30034 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 30035 + }, + { + "epoch": 1.82, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 30036 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 30037 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 30038 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 30039 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 30040 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 30041 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 30042 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1468, + "step": 30043 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.0821, + "step": 30044 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 30045 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 30046 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 30047 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 30048 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 30049 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 30050 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 30051 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 30052 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 30053 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 30054 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1555, + "step": 30055 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 30056 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 30057 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 30058 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 30059 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1765, + "step": 30060 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 30061 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1603, + "step": 30062 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 30063 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 30064 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 30065 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 30066 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 30067 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 30068 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.161, + "step": 30069 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.168, + "step": 30070 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 30071 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 30072 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 30073 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1641, + "step": 30074 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1465, + "step": 30075 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 30076 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 30077 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 30078 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 30079 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 30080 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 30081 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 30082 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 30083 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 30084 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 30085 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1738, + "step": 30086 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 30087 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 30088 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.152, + "step": 30089 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 30090 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 30091 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 30092 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 30093 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 30094 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1513, + "step": 30095 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 30096 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 30097 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 30098 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 30099 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 30100 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 30101 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1627, + "step": 30102 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 30103 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 30104 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 30105 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1533, + "step": 30106 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.164, + "step": 30107 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 30108 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.2117, + "step": 30109 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 30110 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 30111 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1526, + "step": 30112 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 30113 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 30114 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 30115 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 30116 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 30117 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 30118 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1618, + "step": 30119 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 30120 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1759, + "step": 30121 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.0786, + "step": 30122 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1521, + "step": 30123 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 30124 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 30125 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 30126 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1632, + "step": 30127 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 30128 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1756, + "step": 30129 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.159, + "step": 30130 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 30131 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 30132 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 30133 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 30134 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.144, + "step": 30135 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 30136 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 30137 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1816, + "step": 30138 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 30139 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 30140 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.183, + "step": 30141 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 30142 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 30143 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 30144 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1714, + "step": 30145 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 30146 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 30147 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1692, + "step": 30148 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 30149 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.0941, + "step": 30150 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 30151 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 30152 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 30153 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 30154 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 30155 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1541, + "step": 30156 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1627, + "step": 30157 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 30158 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 30159 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 30160 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 30161 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1542, + "step": 30162 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 30163 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 30164 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 30165 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 30166 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 30167 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 30168 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.089, + "step": 30169 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 30170 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 30171 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 30172 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 30173 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 30174 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1538, + "step": 30175 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1468, + "step": 30176 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 30177 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1476, + "step": 30178 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 30179 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1478, + "step": 30180 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1624, + "step": 30181 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 30182 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 30183 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 30184 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 30185 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 30186 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 30187 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 30188 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 30189 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 30190 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 30191 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 30192 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1833, + "step": 30193 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.0764, + "step": 30194 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 30195 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 30196 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 30197 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 30198 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1748, + "step": 30199 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 30200 + }, + { + "epoch": 1.83, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 30201 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 30202 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.0862, + "step": 30203 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.147, + "step": 30204 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1573, + "step": 30205 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 30206 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1478, + "step": 30207 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 30208 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1768, + "step": 30209 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.148, + "step": 30210 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 30211 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 30212 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.167, + "step": 30213 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 30214 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 30215 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 30216 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.0909, + "step": 30217 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 30218 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 30219 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 30220 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 30221 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 30222 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 30223 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1791, + "step": 30224 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 30225 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 30226 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 30227 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 30228 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 30229 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 30230 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 30231 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1657, + "step": 30232 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 30233 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 30234 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 30235 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 30236 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 30237 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 30238 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 30239 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 30240 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 30241 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 30242 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 30243 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 30244 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 30245 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 30246 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 30247 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 30248 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 30249 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 30250 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.17, + "step": 30251 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 30252 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 30253 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 30254 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 30255 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 30256 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 30257 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 30258 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 30259 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 30260 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 30261 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 30262 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.0889, + "step": 30263 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 30264 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 30265 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 30266 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 30267 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 30268 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 30269 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 30270 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 30271 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1535, + "step": 30272 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 30273 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 30274 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 30275 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 30276 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 30277 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 30278 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 30279 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 30280 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 30281 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1716, + "step": 30282 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 30283 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1616, + "step": 30284 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 30285 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 30286 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1532, + "step": 30287 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 30288 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1528, + "step": 30289 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 30290 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 30291 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 30292 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 30293 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 30294 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 30295 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 30296 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 30297 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 30298 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 30299 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1601, + "step": 30300 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.176, + "step": 30301 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 30302 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 30303 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 30304 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 30305 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 30306 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 30307 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 30308 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1468, + "step": 30309 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 30310 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1664, + "step": 30311 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 30312 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 30313 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 30314 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 30315 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 30316 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 30317 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 30318 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1555, + "step": 30319 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 30320 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 30321 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1952, + "step": 30322 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 30323 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1713, + "step": 30324 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 30325 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 30326 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 30327 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 30328 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1554, + "step": 30329 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 30330 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 30331 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1588, + "step": 30332 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 30333 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1546, + "step": 30334 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1685, + "step": 30335 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 30336 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 30337 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 30338 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1918, + "step": 30339 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 30340 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1476, + "step": 30341 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 30342 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 30343 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 30344 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 30345 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 30346 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 30347 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 30348 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 30349 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1686, + "step": 30350 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1511, + "step": 30351 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 30352 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1593, + "step": 30353 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1732, + "step": 30354 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1752, + "step": 30355 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 30356 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 30357 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 30358 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1442, + "step": 30359 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 30360 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 30361 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.152, + "step": 30362 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 30363 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 30364 + }, + { + "epoch": 1.84, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 30365 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 30366 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 30367 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 30368 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 30369 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 30370 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 30371 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 30372 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 30373 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 30374 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 30375 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 30376 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1534, + "step": 30377 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 30378 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 30379 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 30380 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.084, + "step": 30381 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 30382 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 30383 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1579, + "step": 30384 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 30385 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 30386 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 30387 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 30388 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 30389 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 30390 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 30391 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1525, + "step": 30392 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 30393 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1648, + "step": 30394 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 30395 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 30396 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 30397 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 30398 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1563, + "step": 30399 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 30400 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 30401 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 30402 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1629, + "step": 30403 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 30404 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 30405 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 30406 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 30407 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 30408 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 30409 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 30410 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 30411 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 30412 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 30413 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 30414 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1668, + "step": 30415 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1586, + "step": 30416 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 30417 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 30418 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 30419 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 30420 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 30421 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 30422 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.0899, + "step": 30423 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 30424 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 30425 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 30426 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 30427 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1873, + "step": 30428 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 30429 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1691, + "step": 30430 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1531, + "step": 30431 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 30432 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1533, + "step": 30433 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1748, + "step": 30434 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.2045, + "step": 30435 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 30436 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 30437 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 30438 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1807, + "step": 30439 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 30440 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 30441 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 30442 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 30443 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 30444 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.178, + "step": 30445 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1666, + "step": 30446 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1456, + "step": 30447 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 30448 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 30449 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 30450 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 30451 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 30452 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1633, + "step": 30453 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1702, + "step": 30454 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 30455 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 30456 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.164, + "step": 30457 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 30458 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 30459 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 30460 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 30461 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 30462 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 30463 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 30464 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 30465 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 30466 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 30467 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 30468 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 30469 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 30470 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 30471 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.175, + "step": 30472 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 30473 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 30474 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 30475 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 30476 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 30477 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 30478 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 30479 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1542, + "step": 30480 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 30481 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 30482 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1677, + "step": 30483 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 30484 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 30485 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 30486 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 30487 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1654, + "step": 30488 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1611, + "step": 30489 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 30490 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1617, + "step": 30491 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1462, + "step": 30492 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1986, + "step": 30493 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1516, + "step": 30494 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 30495 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1585, + "step": 30496 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 30497 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 30498 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 30499 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 30500 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 30501 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 30502 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 30503 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 30504 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 30505 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1553, + "step": 30506 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 30507 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 30508 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 30509 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1724, + "step": 30510 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 30511 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 30512 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 30513 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 30514 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 30515 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1485, + "step": 30516 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 30517 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 30518 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 30519 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 30520 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 30521 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1694, + "step": 30522 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 30523 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 30524 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 30525 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 30526 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1524, + "step": 30527 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 30528 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 30529 + }, + { + "epoch": 1.85, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 30530 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 30531 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1748, + "step": 30532 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1459, + "step": 30533 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 30534 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 30535 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 30536 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 30537 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 30538 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 30539 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 30540 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 30541 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 30542 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1476, + "step": 30543 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 30544 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.16, + "step": 30545 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 30546 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 30547 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 30548 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1643, + "step": 30549 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 30550 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.164, + "step": 30551 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 30552 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 30553 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1772, + "step": 30554 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 30555 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 30556 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 30557 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 30558 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 30559 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 30560 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 30561 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 30562 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 30563 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 30564 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 30565 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1742, + "step": 30566 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 30567 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 30568 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1709, + "step": 30569 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 30570 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1613, + "step": 30571 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 30572 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 30573 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1798, + "step": 30574 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 30575 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 30576 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1612, + "step": 30577 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1649, + "step": 30578 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 30579 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.171, + "step": 30580 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 30581 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 30582 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1555, + "step": 30583 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 30584 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 30585 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 30586 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 30587 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 30588 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 30589 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 30590 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 30591 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 30592 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 30593 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 30594 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 30595 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 30596 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1534, + "step": 30597 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 30598 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 30599 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 30600 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 30601 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 30602 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1653, + "step": 30603 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 30604 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 30605 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 30606 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 30607 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 30608 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 30609 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 30610 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 30611 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 30612 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1644, + "step": 30613 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1666, + "step": 30614 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 30615 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 30616 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 30617 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1687, + "step": 30618 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 30619 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 30620 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 30621 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 30622 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 30623 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 30624 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 30625 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 30626 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.146, + "step": 30627 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 30628 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 30629 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 30630 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1558, + "step": 30631 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 30632 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 30633 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 30634 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 30635 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 30636 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1832, + "step": 30637 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 30638 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1996, + "step": 30639 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.173, + "step": 30640 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 30641 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1641, + "step": 30642 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 30643 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 30644 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 30645 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 30646 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 30647 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 30648 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 30649 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 30650 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 30651 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 30652 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 30653 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 30654 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 30655 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.2067, + "step": 30656 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 30657 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1593, + "step": 30658 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 30659 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1639, + "step": 30660 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 30661 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1515, + "step": 30662 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 30663 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 30664 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 30665 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1515, + "step": 30666 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 30667 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 30668 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 30669 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.161, + "step": 30670 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 30671 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 30672 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 30673 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 30674 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 30675 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1538, + "step": 30676 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 30677 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 30678 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 30679 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 30680 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 30681 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1538, + "step": 30682 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.186, + "step": 30683 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1613, + "step": 30684 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 30685 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.0886, + "step": 30686 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 30687 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 30688 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 30689 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 30690 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1675, + "step": 30691 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 30692 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 30693 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.165, + "step": 30694 + }, + { + "epoch": 1.86, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 30695 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 30696 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 30697 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1722, + "step": 30698 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 30699 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 30700 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 30701 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1488, + "step": 30702 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 30703 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 30704 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 30705 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 30706 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 30707 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1532, + "step": 30708 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1503, + "step": 30709 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 30710 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1706, + "step": 30711 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 30712 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 30713 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 30714 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1655, + "step": 30715 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 30716 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 30717 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 30718 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 30719 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 30720 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1841, + "step": 30721 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1453, + "step": 30722 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 30723 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 30724 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 30725 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 30726 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 30727 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1602, + "step": 30728 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 30729 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 30730 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 30731 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 30732 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1622, + "step": 30733 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 30734 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 30735 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 30736 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 30737 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 30738 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 30739 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 30740 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 30741 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 30742 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 30743 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 30744 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1526, + "step": 30745 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 30746 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 30747 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 30748 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1704, + "step": 30749 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 30750 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1736, + "step": 30751 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1453, + "step": 30752 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1507, + "step": 30753 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 30754 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 30755 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 30756 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 30757 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 30758 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 30759 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 30760 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 30761 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1776, + "step": 30762 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 30763 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 30764 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 30765 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 30766 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 30767 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 30768 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 30769 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 30770 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 30771 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 30772 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 30773 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 30774 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 30775 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 30776 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1655, + "step": 30777 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 30778 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 30779 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 30780 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 30781 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 30782 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 30783 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1822, + "step": 30784 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1582, + "step": 30785 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1753, + "step": 30786 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 30787 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1589, + "step": 30788 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1526, + "step": 30789 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 30790 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 30791 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 30792 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 30793 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 30794 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 30795 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.2048, + "step": 30796 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1614, + "step": 30797 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 30798 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 30799 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 30800 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 30801 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 30802 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 30803 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1651, + "step": 30804 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1877, + "step": 30805 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 30806 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.0929, + "step": 30807 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 30808 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1979, + "step": 30809 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 30810 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 30811 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 30812 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 30813 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 30814 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 30815 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 30816 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 30817 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 30818 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1526, + "step": 30819 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1488, + "step": 30820 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 30821 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 30822 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 30823 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 30824 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 30825 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 30826 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 30827 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1541, + "step": 30828 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 30829 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 30830 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1527, + "step": 30831 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1588, + "step": 30832 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 30833 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1657, + "step": 30834 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 30835 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 30836 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 30837 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1442, + "step": 30838 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1527, + "step": 30839 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 30840 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.0756, + "step": 30841 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 30842 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 30843 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 30844 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 30845 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 30846 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1684, + "step": 30847 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1748, + "step": 30848 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1731, + "step": 30849 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 30850 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 30851 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1468, + "step": 30852 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 30853 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 30854 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 30855 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1648, + "step": 30856 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 30857 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 30858 + }, + { + "epoch": 1.87, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 30859 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 30860 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 30861 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 30862 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 30863 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1705, + "step": 30864 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 30865 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 30866 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 30867 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 30868 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 30869 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 30870 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 30871 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 30872 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 30873 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 30874 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 30875 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 30876 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 30877 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 30878 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 30879 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 30880 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 30881 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 30882 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 30883 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1516, + "step": 30884 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 30885 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 30886 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 30887 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1671, + "step": 30888 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 30889 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 30890 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 30891 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 30892 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1644, + "step": 30893 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 30894 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 30895 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1623, + "step": 30896 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 30897 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 30898 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 30899 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 30900 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 30901 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 30902 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 30903 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 30904 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 30905 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1654, + "step": 30906 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 30907 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 30908 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 30909 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 30910 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 30911 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1657, + "step": 30912 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 30913 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 30914 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 30915 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 30916 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 30917 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 30918 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 30919 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1567, + "step": 30920 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 30921 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 30922 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 30923 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 30924 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1495, + "step": 30925 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 30926 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 30927 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 30928 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 30929 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 30930 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 30931 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1511, + "step": 30932 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 30933 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1812, + "step": 30934 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 30935 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1429, + "step": 30936 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1507, + "step": 30937 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1527, + "step": 30938 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1499, + "step": 30939 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 30940 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 30941 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 30942 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.178, + "step": 30943 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 30944 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 30945 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1892, + "step": 30946 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 30947 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1513, + "step": 30948 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 30949 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 30950 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 30951 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 30952 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 30953 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 30954 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 30955 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 30956 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 30957 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 30958 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 30959 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 30960 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 30961 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 30962 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 30963 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 30964 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1513, + "step": 30965 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 30966 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 30967 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 30968 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 30969 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 30970 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 30971 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 30972 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 30973 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 30974 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 30975 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 30976 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 30977 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1511, + "step": 30978 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 30979 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1517, + "step": 30980 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 30981 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 30982 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 30983 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 30984 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 30985 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 30986 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 30987 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 30988 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 30989 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1598, + "step": 30990 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1555, + "step": 30991 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 30992 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 30993 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 30994 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 30995 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 30996 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1631, + "step": 30997 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1685, + "step": 30998 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.152, + "step": 30999 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 31000 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 31001 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 31002 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1821, + "step": 31003 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 31004 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 31005 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 31006 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 31007 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 31008 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 31009 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 31010 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 31011 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 31012 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1546, + "step": 31013 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.0941, + "step": 31014 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 31015 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 31016 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 31017 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 31018 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 31019 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 31020 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 31021 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 31022 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1656, + "step": 31023 + }, + { + "epoch": 1.88, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 31024 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 31025 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 31026 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 31027 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 31028 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 31029 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 31030 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 31031 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 31032 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1757, + "step": 31033 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1663, + "step": 31034 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1622, + "step": 31035 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 31036 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 31037 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1618, + "step": 31038 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 31039 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1939, + "step": 31040 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.145, + "step": 31041 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 31042 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 31043 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 31044 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 31045 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 31046 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1645, + "step": 31047 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 31048 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 31049 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1642, + "step": 31050 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 31051 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1713, + "step": 31052 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 31053 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 31054 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1525, + "step": 31055 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.158, + "step": 31056 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 31057 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 31058 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 31059 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1657, + "step": 31060 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 31061 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 31062 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 31063 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1738, + "step": 31064 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 31065 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1585, + "step": 31066 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 31067 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 31068 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 31069 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 31070 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 31071 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 31072 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 31073 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 31074 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 31075 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1765, + "step": 31076 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 31077 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1622, + "step": 31078 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1515, + "step": 31079 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1664, + "step": 31080 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 31081 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 31082 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 31083 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 31084 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1568, + "step": 31085 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1534, + "step": 31086 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1733, + "step": 31087 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 31088 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1749, + "step": 31089 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 31090 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 31091 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 31092 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1739, + "step": 31093 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1633, + "step": 31094 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 31095 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1752, + "step": 31096 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 31097 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 31098 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1617, + "step": 31099 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 31100 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 31101 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 31102 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 31103 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 31104 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 31105 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 31106 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 31107 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 31108 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 31109 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 31110 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 31111 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 31112 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 31113 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 31114 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 31115 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1521, + "step": 31116 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 31117 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1586, + "step": 31118 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 31119 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 31120 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1656, + "step": 31121 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 31122 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.0911, + "step": 31123 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1429, + "step": 31124 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 31125 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 31126 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 31127 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1739, + "step": 31128 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 31129 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 31130 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1649, + "step": 31131 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 31132 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 31133 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 31134 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 31135 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 31136 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 31137 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 31138 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1679, + "step": 31139 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 31140 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 31141 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 31142 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 31143 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 31144 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 31145 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 31146 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1761, + "step": 31147 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 31148 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 31149 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 31150 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 31151 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1776, + "step": 31152 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 31153 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 31154 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1612, + "step": 31155 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 31156 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 31157 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 31158 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1589, + "step": 31159 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 31160 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 31161 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 31162 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 31163 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 31164 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 31165 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 31166 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.146, + "step": 31167 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 31168 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 31169 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1658, + "step": 31170 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 31171 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1644, + "step": 31172 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 31173 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1588, + "step": 31174 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1498, + "step": 31175 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 31176 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1714, + "step": 31177 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 31178 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 31179 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1555, + "step": 31180 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 31181 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1762, + "step": 31182 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 31183 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 31184 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 31185 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 31186 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 31187 + }, + { + "epoch": 1.89, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 31188 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 31189 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 31190 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.153, + "step": 31191 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1534, + "step": 31192 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1648, + "step": 31193 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 31194 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 31195 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1825, + "step": 31196 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 31197 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 31198 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1801, + "step": 31199 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 31200 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1507, + "step": 31201 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 31202 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 31203 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 31204 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 31205 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 31206 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 31207 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 31208 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 31209 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 31210 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 31211 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 31212 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 31213 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1552, + "step": 31214 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 31215 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1605, + "step": 31216 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 31217 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 31218 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 31219 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 31220 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 31221 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 31222 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 31223 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 31224 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 31225 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 31226 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 31227 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 31228 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 31229 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 31230 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 31231 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 31232 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 31233 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1586, + "step": 31234 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 31235 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 31236 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 31237 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 31238 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.172, + "step": 31239 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 31240 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 31241 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 31242 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 31243 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 31244 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.0906, + "step": 31245 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1772, + "step": 31246 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 31247 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 31248 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 31249 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.2072, + "step": 31250 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.2384, + "step": 31251 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 31252 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 31253 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 31254 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 31255 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 31256 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 31257 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1551, + "step": 31258 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.146, + "step": 31259 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1478, + "step": 31260 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 31261 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 31262 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 31263 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 31264 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 31265 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 31266 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 31267 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1628, + "step": 31268 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 31269 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 31270 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 31271 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 31272 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1532, + "step": 31273 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 31274 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 31275 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 31276 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 31277 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1599, + "step": 31278 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 31279 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 31280 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 31281 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 31282 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 31283 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 31284 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1588, + "step": 31285 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 31286 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1546, + "step": 31287 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 31288 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 31289 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 31290 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1549, + "step": 31291 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 31292 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 31293 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 31294 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 31295 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 31296 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 31297 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 31298 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 31299 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1715, + "step": 31300 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 31301 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 31302 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1606, + "step": 31303 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 31304 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 31305 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 31306 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 31307 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1672, + "step": 31308 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 31309 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1548, + "step": 31310 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 31311 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 31312 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 31313 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 31314 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 31315 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 31316 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 31317 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1485, + "step": 31318 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1692, + "step": 31319 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 31320 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 31321 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1818, + "step": 31322 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 31323 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 31324 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 31325 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 31326 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1557, + "step": 31327 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 31328 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1806, + "step": 31329 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1692, + "step": 31330 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 31331 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.177, + "step": 31332 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 31333 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 31334 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 31335 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 31336 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 31337 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 31338 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.0845, + "step": 31339 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1612, + "step": 31340 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 31341 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 31342 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 31343 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 31344 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 31345 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 31346 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 31347 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 31348 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 31349 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 31350 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1702, + "step": 31351 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 31352 + }, + { + "epoch": 1.9, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 31353 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 31354 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 31355 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 31356 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.17, + "step": 31357 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 31358 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1848, + "step": 31359 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 31360 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 31361 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 31362 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 31363 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 31364 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1685, + "step": 31365 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 31366 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1983, + "step": 31367 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 31368 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 31369 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 31370 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1589, + "step": 31371 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 31372 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 31373 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 31374 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 31375 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 31376 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 31377 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 31378 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 31379 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1771, + "step": 31380 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 31381 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 31382 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 31383 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1528, + "step": 31384 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 31385 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 31386 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 31387 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 31388 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 31389 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 31390 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 31391 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 31392 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1542, + "step": 31393 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 31394 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1626, + "step": 31395 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 31396 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 31397 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1787, + "step": 31398 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 31399 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 31400 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 31401 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 31402 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1635, + "step": 31403 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 31404 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 31405 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 31406 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 31407 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 31408 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 31409 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 31410 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 31411 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 31412 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 31413 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 31414 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 31415 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 31416 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 31417 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.159, + "step": 31418 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 31419 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.0848, + "step": 31420 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 31421 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 31422 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 31423 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 31424 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 31425 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 31426 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 31427 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 31428 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 31429 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 31430 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 31431 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 31432 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 31433 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 31434 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1538, + "step": 31435 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 31436 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 31437 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1837, + "step": 31438 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 31439 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 31440 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 31441 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.0918, + "step": 31442 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 31443 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 31444 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 31445 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 31446 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 31447 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 31448 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 31449 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 31450 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 31451 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 31452 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1787, + "step": 31453 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 31454 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 31455 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 31456 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 31457 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 31458 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 31459 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 31460 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1511, + "step": 31461 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 31462 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 31463 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 31464 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1912, + "step": 31465 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 31466 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 31467 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 31468 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 31469 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 31470 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1456, + "step": 31471 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 31472 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 31473 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 31474 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 31475 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1507, + "step": 31476 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 31477 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 31478 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 31479 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 31480 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 31481 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 31482 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 31483 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 31484 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 31485 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 31486 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 31487 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1834, + "step": 31488 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1885, + "step": 31489 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.0859, + "step": 31490 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 31491 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 31492 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 31493 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 31494 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1661, + "step": 31495 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 31496 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1844, + "step": 31497 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 31498 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1561, + "step": 31499 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 31500 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 31501 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 31502 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 31503 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.0879, + "step": 31504 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1465, + "step": 31505 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 31506 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 31507 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 31508 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1533, + "step": 31509 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 31510 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 31511 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 31512 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1586, + "step": 31513 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 31514 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 31515 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1867, + "step": 31516 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.0801, + "step": 31517 + }, + { + "epoch": 1.91, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 31518 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 31519 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 31520 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1785, + "step": 31521 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 31522 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 31523 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 31524 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 31525 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.154, + "step": 31526 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1973, + "step": 31527 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 31528 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 31529 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 31530 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.147, + "step": 31531 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 31532 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1531, + "step": 31533 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 31534 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 31535 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 31536 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 31537 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 31538 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 31539 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 31540 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 31541 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 31542 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 31543 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 31544 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 31545 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 31546 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 31547 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 31548 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 31549 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 31550 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 31551 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 31552 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 31553 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 31554 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 31555 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1631, + "step": 31556 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 31557 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 31558 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 31559 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 31560 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.0715, + "step": 31561 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 31562 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1459, + "step": 31563 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 31564 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 31565 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 31566 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1624, + "step": 31567 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 31568 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 31569 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 31570 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1614, + "step": 31571 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 31572 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 31573 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 31574 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 31575 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1752, + "step": 31576 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 31577 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 31578 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1825, + "step": 31579 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1515, + "step": 31580 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 31581 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 31582 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 31583 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 31584 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1841, + "step": 31585 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 31586 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.2138, + "step": 31587 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 31588 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 31589 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1442, + "step": 31590 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 31591 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 31592 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1731, + "step": 31593 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 31594 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 31595 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 31596 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 31597 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.2144, + "step": 31598 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 31599 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 31600 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1642, + "step": 31601 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 31602 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1923, + "step": 31603 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 31604 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 31605 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 31606 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.189, + "step": 31607 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 31608 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 31609 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 31610 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 31611 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 31612 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 31613 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 31614 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 31615 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 31616 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 31617 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 31618 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1588, + "step": 31619 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 31620 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 31621 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 31622 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 31623 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1451, + "step": 31624 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 31625 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 31626 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1808, + "step": 31627 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 31628 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1692, + "step": 31629 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 31630 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 31631 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 31632 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 31633 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 31634 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1749, + "step": 31635 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 31636 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1558, + "step": 31637 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 31638 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1651, + "step": 31639 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 31640 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1603, + "step": 31641 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 31642 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 31643 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 31644 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 31645 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 31646 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 31647 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1561, + "step": 31648 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 31649 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 31650 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 31651 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1451, + "step": 31652 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 31653 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 31654 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 31655 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1583, + "step": 31656 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 31657 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 31658 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1564, + "step": 31659 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1603, + "step": 31660 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 31661 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 31662 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 31663 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 31664 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 31665 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1625, + "step": 31666 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 31667 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 31668 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 31669 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1662, + "step": 31670 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 31671 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 31672 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 31673 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1788, + "step": 31674 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 31675 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 31676 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1733, + "step": 31677 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 31678 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 31679 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 31680 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 31681 + }, + { + "epoch": 1.92, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 31682 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 31683 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 31684 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1478, + "step": 31685 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 31686 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 31687 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1807, + "step": 31688 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1765, + "step": 31689 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1488, + "step": 31690 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 31691 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 31692 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1485, + "step": 31693 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 31694 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 31695 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 31696 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 31697 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 31698 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.153, + "step": 31699 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 31700 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 31701 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 31702 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 31703 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 31704 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 31705 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 31706 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 31707 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 31708 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 31709 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 31710 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 31711 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 31712 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 31713 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 31714 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 31715 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 31716 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 31717 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 31718 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 31719 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 31720 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 31721 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1429, + "step": 31722 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 31723 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 31724 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 31725 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 31726 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 31727 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 31728 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 31729 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 31730 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1527, + "step": 31731 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1664, + "step": 31732 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 31733 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 31734 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 31735 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 31736 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 31737 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 31738 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 31739 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 31740 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1608, + "step": 31741 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 31742 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 31743 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 31744 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 31745 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.155, + "step": 31746 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 31747 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.148, + "step": 31748 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 31749 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 31750 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 31751 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 31752 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 31753 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 31754 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 31755 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 31756 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 31757 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 31758 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 31759 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1612, + "step": 31760 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 31761 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 31762 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 31763 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 31764 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 31765 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 31766 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 31767 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 31768 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1961, + "step": 31769 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 31770 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 31771 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1429, + "step": 31772 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1716, + "step": 31773 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 31774 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 31775 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 31776 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 31777 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.153, + "step": 31778 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1525, + "step": 31779 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 31780 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 31781 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1618, + "step": 31782 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.0899, + "step": 31783 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 31784 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1592, + "step": 31785 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 31786 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 31787 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 31788 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 31789 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 31790 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1606, + "step": 31791 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 31792 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 31793 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 31794 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 31795 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1553, + "step": 31796 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 31797 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 31798 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 31799 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1598, + "step": 31800 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1854, + "step": 31801 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 31802 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 31803 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 31804 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 31805 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1542, + "step": 31806 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 31807 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 31808 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.0823, + "step": 31809 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 31810 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 31811 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 31812 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 31813 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 31814 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1589, + "step": 31815 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 31816 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 31817 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 31818 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 31819 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 31820 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 31821 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 31822 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 31823 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 31824 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 31825 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 31826 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 31827 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 31828 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 31829 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 31830 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 31831 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 31832 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1674, + "step": 31833 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 31834 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 31835 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1597, + "step": 31836 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 31837 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1763, + "step": 31838 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 31839 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.158, + "step": 31840 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 31841 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 31842 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 31843 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 31844 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 31845 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 31846 + }, + { + "epoch": 1.93, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 31847 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 31848 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1605, + "step": 31849 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 31850 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 31851 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 31852 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 31853 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1752, + "step": 31854 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 31855 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1524, + "step": 31856 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 31857 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1497, + "step": 31858 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 31859 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 31860 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1706, + "step": 31861 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 31862 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 31863 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 31864 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 31865 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1533, + "step": 31866 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 31867 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 31868 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 31869 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 31870 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 31871 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 31872 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.152, + "step": 31873 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 31874 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 31875 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 31876 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 31877 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 31878 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 31879 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 31880 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 31881 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 31882 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 31883 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 31884 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1778, + "step": 31885 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 31886 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1548, + "step": 31887 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1532, + "step": 31888 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 31889 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 31890 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1618, + "step": 31891 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 31892 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1661, + "step": 31893 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 31894 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 31895 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 31896 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 31897 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 31898 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 31899 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 31900 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 31901 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1549, + "step": 31902 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1624, + "step": 31903 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 31904 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1599, + "step": 31905 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1626, + "step": 31906 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 31907 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 31908 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 31909 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 31910 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 31911 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 31912 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 31913 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 31914 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 31915 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 31916 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.2072, + "step": 31917 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 31918 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 31919 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 31920 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1813, + "step": 31921 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 31922 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1884, + "step": 31923 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 31924 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 31925 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 31926 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 31927 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 31928 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 31929 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 31930 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 31931 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 31932 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 31933 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1554, + "step": 31934 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 31935 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 31936 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 31937 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 31938 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 31939 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 31940 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 31941 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 31942 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 31943 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 31944 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 31945 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 31946 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 31947 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 31948 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 31949 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 31950 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 31951 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1734, + "step": 31952 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.175, + "step": 31953 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1495, + "step": 31954 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 31955 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 31956 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 31957 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 31958 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 31959 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1507, + "step": 31960 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 31961 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 31962 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1591, + "step": 31963 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1563, + "step": 31964 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 31965 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 31966 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 31967 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 31968 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 31969 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 31970 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 31971 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 31972 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.0891, + "step": 31973 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 31974 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 31975 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 31976 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 31977 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 31978 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 31979 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 31980 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 31981 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 31982 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 31983 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.171, + "step": 31984 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 31985 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 31986 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1788, + "step": 31987 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 31988 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1642, + "step": 31989 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 31990 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 31991 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 31992 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1468, + "step": 31993 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 31994 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 31995 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 31996 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 31997 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 31998 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 31999 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1604, + "step": 32000 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 32001 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1518, + "step": 32002 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1681, + "step": 32003 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 32004 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 32005 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 32006 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 32007 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 32008 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 32009 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1459, + "step": 32010 + }, + { + "epoch": 1.94, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 32011 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 32012 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1664, + "step": 32013 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 32014 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 32015 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 32016 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 32017 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 32018 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 32019 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 32020 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 32021 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 32022 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 32023 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1533, + "step": 32024 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 32025 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 32026 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 32027 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 32028 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 32029 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 32030 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 32031 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 32032 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 32033 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 32034 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 32035 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 32036 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 32037 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 32038 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1561, + "step": 32039 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 32040 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 32041 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 32042 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 32043 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 32044 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1602, + "step": 32045 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.162, + "step": 32046 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 32047 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.0874, + "step": 32048 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 32049 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 32050 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.08, + "step": 32051 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 32052 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 32053 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 32054 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 32055 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1635, + "step": 32056 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 32057 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 32058 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 32059 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1674, + "step": 32060 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1531, + "step": 32061 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 32062 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 32063 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 32064 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1694, + "step": 32065 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 32066 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1761, + "step": 32067 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1516, + "step": 32068 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 32069 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 32070 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 32071 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 32072 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 32073 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1456, + "step": 32074 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 32075 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 32076 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 32077 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 32078 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 32079 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1637, + "step": 32080 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 32081 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 32082 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 32083 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 32084 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 32085 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 32086 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 32087 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 32088 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 32089 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 32090 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 32091 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 32092 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 32093 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 32094 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 32095 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 32096 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1778, + "step": 32097 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 32098 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 32099 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1564, + "step": 32100 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 32101 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 32102 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 32103 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.0941, + "step": 32104 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1552, + "step": 32105 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 32106 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 32107 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 32108 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 32109 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 32110 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1901, + "step": 32111 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 32112 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 32113 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 32114 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1617, + "step": 32115 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 32116 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 32117 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 32118 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1665, + "step": 32119 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 32120 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 32121 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 32122 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 32123 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 32124 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 32125 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 32126 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 32127 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1663, + "step": 32128 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 32129 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1586, + "step": 32130 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 32131 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 32132 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1676, + "step": 32133 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 32134 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 32135 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 32136 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 32137 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 32138 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 32139 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 32140 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 32141 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 32142 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 32143 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 32144 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 32145 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 32146 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 32147 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 32148 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 32149 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 32150 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 32151 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 32152 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 32153 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 32154 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 32155 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 32156 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 32157 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 32158 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 32159 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 32160 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 32161 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 32162 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 32163 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1926, + "step": 32164 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 32165 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 32166 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 32167 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 32168 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1507, + "step": 32169 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 32170 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 32171 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 32172 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 32173 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 32174 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 32175 + }, + { + "epoch": 1.95, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 32176 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 32177 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 32178 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 32179 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 32180 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 32181 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 32182 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 32183 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 32184 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 32185 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 32186 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 32187 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 32188 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 32189 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 32190 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 32191 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 32192 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 32193 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 32194 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1841, + "step": 32195 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 32196 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 32197 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 32198 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 32199 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 32200 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 32201 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.2018, + "step": 32202 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 32203 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.148, + "step": 32204 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1593, + "step": 32205 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 32206 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1563, + "step": 32207 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 32208 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 32209 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.178, + "step": 32210 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 32211 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 32212 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1521, + "step": 32213 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 32214 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1567, + "step": 32215 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 32216 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 32217 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 32218 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 32219 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 32220 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 32221 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 32222 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1602, + "step": 32223 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 32224 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 32225 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 32226 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 32227 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 32228 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 32229 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 32230 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 32231 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 32232 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 32233 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 32234 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 32235 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 32236 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 32237 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1605, + "step": 32238 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 32239 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 32240 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1573, + "step": 32241 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.0901, + "step": 32242 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 32243 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 32244 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1991, + "step": 32245 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 32246 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 32247 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 32248 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1617, + "step": 32249 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 32250 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1488, + "step": 32251 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 32252 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 32253 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 32254 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 32255 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 32256 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 32257 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 32258 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 32259 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1657, + "step": 32260 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 32261 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1533, + "step": 32262 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 32263 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 32264 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 32265 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 32266 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 32267 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1597, + "step": 32268 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 32269 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 32270 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1558, + "step": 32271 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 32272 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 32273 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 32274 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 32275 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 32276 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 32277 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 32278 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 32279 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 32280 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 32281 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 32282 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 32283 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 32284 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 32285 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 32286 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 32287 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1521, + "step": 32288 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 32289 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 32290 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 32291 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 32292 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 32293 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 32294 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 32295 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.0929, + "step": 32296 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 32297 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 32298 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 32299 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1791, + "step": 32300 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.0906, + "step": 32301 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 32302 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 32303 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 32304 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 32305 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 32306 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 32307 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 32308 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 32309 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 32310 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 32311 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1785, + "step": 32312 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1694, + "step": 32313 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 32314 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1805, + "step": 32315 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 32316 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 32317 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 32318 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 32319 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 32320 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 32321 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 32322 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1811, + "step": 32323 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 32324 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 32325 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 32326 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1883, + "step": 32327 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1754, + "step": 32328 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1531, + "step": 32329 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 32330 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1674, + "step": 32331 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1698, + "step": 32332 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 32333 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 32334 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 32335 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 32336 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 32337 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 32338 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 32339 + }, + { + "epoch": 1.96, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 32340 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 32341 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 32342 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1769, + "step": 32343 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1518, + "step": 32344 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 32345 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.0866, + "step": 32346 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 32347 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 32348 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 32349 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 32350 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 32351 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 32352 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 32353 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 32354 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1535, + "step": 32355 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 32356 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.158, + "step": 32357 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 32358 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 32359 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 32360 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1562, + "step": 32361 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 32362 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 32363 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1552, + "step": 32364 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 32365 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 32366 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 32367 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 32368 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 32369 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 32370 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1709, + "step": 32371 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 32372 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1663, + "step": 32373 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 32374 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 32375 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 32376 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1669, + "step": 32377 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 32378 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 32379 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 32380 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 32381 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1503, + "step": 32382 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 32383 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 32384 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1542, + "step": 32385 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 32386 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 32387 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 32388 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 32389 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 32390 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 32391 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 32392 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 32393 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 32394 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1524, + "step": 32395 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 32396 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 32397 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 32398 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1647, + "step": 32399 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 32400 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 32401 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 32402 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 32403 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.2105, + "step": 32404 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 32405 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 32406 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 32407 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 32408 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 32409 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 32410 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 32411 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 32412 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 32413 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 32414 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 32415 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 32416 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 32417 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 32418 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1617, + "step": 32419 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 32420 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1661, + "step": 32421 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1611, + "step": 32422 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.0855, + "step": 32423 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1517, + "step": 32424 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 32425 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 32426 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 32427 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1729, + "step": 32428 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 32429 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 32430 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 32431 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 32432 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1524, + "step": 32433 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 32434 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 32435 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 32436 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1552, + "step": 32437 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 32438 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 32439 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 32440 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 32441 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1535, + "step": 32442 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 32443 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 32444 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1747, + "step": 32445 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 32446 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 32447 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 32448 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 32449 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.0902, + "step": 32450 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 32451 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1676, + "step": 32452 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 32453 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 32454 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 32455 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 32456 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 32457 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 32458 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1548, + "step": 32459 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1564, + "step": 32460 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 32461 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1699, + "step": 32462 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 32463 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 32464 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1825, + "step": 32465 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 32466 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 32467 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 32468 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 32469 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 32470 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1958, + "step": 32471 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 32472 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1555, + "step": 32473 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 32474 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1793, + "step": 32475 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 32476 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 32477 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 32478 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 32479 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1593, + "step": 32480 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 32481 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 32482 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 32483 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 32484 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1518, + "step": 32485 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 32486 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1867, + "step": 32487 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 32488 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1659, + "step": 32489 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 32490 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 32491 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 32492 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 32493 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 32494 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 32495 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 32496 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1443, + "step": 32497 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1738, + "step": 32498 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 32499 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 32500 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 32501 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 32502 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 32503 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.1564, + "step": 32504 + }, + { + "epoch": 1.97, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 32505 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 32506 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 32507 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 32508 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.144, + "step": 32509 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 32510 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 32511 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 32512 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.0882, + "step": 32513 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 32514 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1647, + "step": 32515 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1526, + "step": 32516 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 32517 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 32518 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1644, + "step": 32519 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 32520 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 32521 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 32522 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1603, + "step": 32523 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 32524 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1942, + "step": 32525 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 32526 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 32527 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 32528 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 32529 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 32530 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 32531 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1597, + "step": 32532 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1527, + "step": 32533 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 32534 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 32535 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 32536 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 32537 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 32538 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 32539 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 32540 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 32541 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1548, + "step": 32542 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 32543 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 32544 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 32545 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 32546 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 32547 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 32548 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 32549 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 32550 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 32551 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 32552 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1634, + "step": 32553 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 32554 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 32555 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1797, + "step": 32556 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 32557 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 32558 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1754, + "step": 32559 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 32560 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 32561 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 32562 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 32563 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 32564 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1623, + "step": 32565 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 32566 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 32567 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1602, + "step": 32568 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1604, + "step": 32569 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 32570 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 32571 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 32572 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 32573 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 32574 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1621, + "step": 32575 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 32576 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1632, + "step": 32577 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1687, + "step": 32578 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 32579 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 32580 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 32581 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1676, + "step": 32582 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 32583 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 32584 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 32585 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 32586 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 32587 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.169, + "step": 32588 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1555, + "step": 32589 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 32590 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1652, + "step": 32591 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 32592 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 32593 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 32594 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 32595 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 32596 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1478, + "step": 32597 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.155, + "step": 32598 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 32599 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.144, + "step": 32600 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1453, + "step": 32601 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 32602 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 32603 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 32604 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 32605 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1704, + "step": 32606 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 32607 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 32608 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 32609 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 32610 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1934, + "step": 32611 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 32612 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 32613 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 32614 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 32615 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.187, + "step": 32616 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 32617 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 32618 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 32619 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 32620 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 32621 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1503, + "step": 32622 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 32623 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 32624 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1836, + "step": 32625 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 32626 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 32627 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 32628 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 32629 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 32630 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 32631 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 32632 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 32633 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 32634 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1574, + "step": 32635 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.147, + "step": 32636 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 32637 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 32638 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 32639 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 32640 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 32641 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 32642 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1521, + "step": 32643 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1691, + "step": 32644 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 32645 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 32646 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1862, + "step": 32647 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.186, + "step": 32648 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 32649 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 32650 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 32651 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 32652 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1513, + "step": 32653 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 32654 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 32655 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1469, + "step": 32656 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1499, + "step": 32657 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 32658 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 32659 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1673, + "step": 32660 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 32661 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1548, + "step": 32662 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 32663 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 32664 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 32665 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 32666 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 32667 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 32668 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 32669 + }, + { + "epoch": 1.98, + "learning_rate": 2e-05, + "loss": 0.1622, + "step": 32670 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 32671 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 32672 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 32673 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 32674 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1507, + "step": 32675 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 32676 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 32677 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 32678 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 32679 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 32680 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1695, + "step": 32681 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 32682 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 32683 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 32684 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1576, + "step": 32685 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 32686 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1497, + "step": 32687 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1534, + "step": 32688 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 32689 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 32690 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 32691 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 32692 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 32693 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 32694 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 32695 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 32696 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 32697 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 32698 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 32699 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 32700 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 32701 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 32702 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 32703 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 32704 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 32705 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 32706 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 32707 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.154, + "step": 32708 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1586, + "step": 32709 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 32710 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1631, + "step": 32711 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1676, + "step": 32712 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 32713 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 32714 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 32715 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 32716 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 32717 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1596, + "step": 32718 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 32719 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 32720 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 32721 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 32722 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 32723 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 32724 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 32725 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 32726 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 32727 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 32728 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 32729 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 32730 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 32731 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 32732 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 32733 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 32734 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1893, + "step": 32735 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 32736 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 32737 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 32738 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1684, + "step": 32739 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 32740 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 32741 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 32742 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 32743 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 32744 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 32745 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1634, + "step": 32746 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 32747 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 32748 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 32749 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 32750 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 32751 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 32752 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 32753 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.159, + "step": 32754 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 32755 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 32756 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 32757 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 32758 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 32759 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 32760 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 32761 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 32762 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1599, + "step": 32763 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 32764 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 32765 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1631, + "step": 32766 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1728, + "step": 32767 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1516, + "step": 32768 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 32769 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 32770 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1591, + "step": 32771 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 32772 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 32773 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 32774 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 32775 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 32776 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 32777 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 32778 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1805, + "step": 32779 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 32780 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 32781 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 32782 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 32783 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1515, + "step": 32784 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 32785 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 32786 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 32787 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 32788 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 32789 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 32790 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 32791 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 32792 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 32793 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 32794 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 32795 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 32796 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 32797 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 32798 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1566, + "step": 32799 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 32800 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 32801 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1692, + "step": 32802 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 32803 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 32804 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 32805 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 32806 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 32807 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 32808 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 32809 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1577, + "step": 32810 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 32811 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 32812 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 32813 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1611, + "step": 32814 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 32815 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 32816 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 32817 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 32818 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 32819 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 32820 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1613, + "step": 32821 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 32822 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 32823 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 32824 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1938, + "step": 32825 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 32826 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 32827 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 32828 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 32829 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 32830 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1498, + "step": 32831 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 32832 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 32833 + }, + { + "epoch": 1.99, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 32834 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 32835 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 32836 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 32837 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1658, + "step": 32838 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 32839 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 32840 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 32841 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 32842 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 32843 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 32844 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 32845 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1637, + "step": 32846 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 32847 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1755, + "step": 32848 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 32849 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 32850 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 32851 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 32852 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 32853 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 32854 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 32855 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 32856 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1791, + "step": 32857 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.0941, + "step": 32858 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 32859 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 32860 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 32861 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 32862 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 32863 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 32864 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 32865 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 32866 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 32867 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1643, + "step": 32868 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 32869 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 32870 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.161, + "step": 32871 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 32872 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 32873 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 32874 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.152, + "step": 32875 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1429, + "step": 32876 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 32877 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1459, + "step": 32878 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 32879 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 32880 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 32881 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 32882 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 32883 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 32884 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 32885 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1637, + "step": 32886 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.163, + "step": 32887 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 32888 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 32889 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 32890 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 32891 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1845, + "step": 32892 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 32893 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 32894 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 32895 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 32896 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 32897 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 32898 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.168, + "step": 32899 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 32900 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1684, + "step": 32901 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1824, + "step": 32902 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 32903 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 32904 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 32905 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 32906 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.165, + "step": 32907 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 32908 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 32909 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1447, + "step": 32910 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 32911 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 32912 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1451, + "step": 32913 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 32914 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 32915 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 32916 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 32917 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.0839, + "step": 32918 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 32919 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 32920 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 32921 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 32922 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 32923 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 32924 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 32925 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.0838, + "step": 32926 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 32927 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 32928 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 32929 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 32930 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 32931 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 32932 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 32933 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 32934 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 32935 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 32936 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 32937 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 32938 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 32939 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 32940 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 32941 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 32942 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 32943 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 32944 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 32945 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 32946 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 32947 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 32948 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 32949 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 32950 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 32951 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 32952 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.0789, + "step": 32953 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 32954 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 32955 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 32956 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 32957 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 32958 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 32959 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.0761, + "step": 32960 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 32961 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 32962 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 32963 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 32964 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.0802, + "step": 32965 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 32966 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1779, + "step": 32967 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.0872, + "step": 32968 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 32969 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.0935, + "step": 32970 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 32971 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 32972 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 32973 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 32974 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 32975 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 32976 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 32977 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 32978 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 32979 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 32980 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 32981 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 32982 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 32983 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 32984 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 32985 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 32986 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 32987 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1497, + "step": 32988 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 32989 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.0881, + "step": 32990 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 32991 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.0864, + "step": 32992 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 32993 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 32994 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 32995 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 32996 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 32997 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 32998 + }, + { + "epoch": 2.0, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 32999 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 33000 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 33001 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 33002 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0769, + "step": 33003 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 33004 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 33005 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 33006 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0847, + "step": 33007 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 33008 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 33009 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 33010 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 33011 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 33012 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 33013 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 33014 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 33015 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 33016 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 33017 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0836, + "step": 33018 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 33019 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 33020 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 33021 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 33022 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 33023 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 33024 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 33025 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 33026 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 33027 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 33028 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0897, + "step": 33029 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0836, + "step": 33030 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 33031 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 33032 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 33033 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 33034 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 33035 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1519, + "step": 33036 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1579, + "step": 33037 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 33038 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 33039 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 33040 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 33041 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 33042 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 33043 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 33044 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 33045 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 33046 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 33047 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 33048 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 33049 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 33050 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 33051 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 33052 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 33053 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 33054 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 33055 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 33056 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 33057 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0808, + "step": 33058 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 33059 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 33060 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 33061 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0906, + "step": 33062 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 33063 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 33064 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 33065 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0843, + "step": 33066 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 33067 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0881, + "step": 33068 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 33069 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 33070 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 33071 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.147, + "step": 33072 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 33073 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 33074 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 33075 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 33076 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 33077 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 33078 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 33079 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1591, + "step": 33080 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 33081 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 33082 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 33083 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 33084 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 33085 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 33086 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 33087 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 33088 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 33089 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 33090 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 33091 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0827, + "step": 33092 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 33093 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 33094 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0785, + "step": 33095 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 33096 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 33097 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 33098 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0806, + "step": 33099 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 33100 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 33101 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 33102 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 33103 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 33104 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 33105 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0865, + "step": 33106 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 33107 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 33108 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0853, + "step": 33109 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1553, + "step": 33110 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0838, + "step": 33111 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0914, + "step": 33112 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0655, + "step": 33113 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 33114 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 33115 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 33116 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 33117 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 33118 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 33119 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 33120 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 33121 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 33122 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 33123 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 33124 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 33125 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 33126 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 33127 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 33128 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 33129 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 33130 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 33131 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 33132 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 33133 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 33134 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 33135 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 33136 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0897, + "step": 33137 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 33138 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 33139 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 33140 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 33141 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 33142 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 33143 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 33144 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 33145 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0787, + "step": 33146 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 33147 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 33148 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0762, + "step": 33149 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 33150 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 33151 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 33152 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0932, + "step": 33153 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 33154 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 33155 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0874, + "step": 33156 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 33157 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 33158 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 33159 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 33160 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 33161 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 33162 + }, + { + "epoch": 2.01, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 33163 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 33164 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 33165 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 33166 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 33167 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 33168 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 33169 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 33170 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.0864, + "step": 33171 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 33172 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 33173 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 33174 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 33175 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 33176 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 33177 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 33178 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 33179 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 33180 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 33181 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 33182 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 33183 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 33184 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.0875, + "step": 33185 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 33186 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 33187 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 33188 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.0859, + "step": 33189 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 33190 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 33191 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.0892, + "step": 33192 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 33193 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 33194 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 33195 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 33196 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 33197 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 33198 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 33199 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 33200 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 33201 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 33202 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 33203 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 33204 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 33205 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 33206 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 33207 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 33208 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 33209 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 33210 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 33211 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.091, + "step": 33212 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.0877, + "step": 33213 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 33214 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 33215 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 33216 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 33217 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 33218 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 33219 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 33220 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 33221 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 33222 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 33223 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 33224 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 33225 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 33226 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 33227 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 33228 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 33229 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 33230 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 33231 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 33232 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 33233 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 33234 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 33235 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 33236 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 33237 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1638, + "step": 33238 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 33239 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 33240 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 33241 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 33242 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 33243 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 33244 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 33245 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 33246 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 33247 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 33248 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 33249 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 33250 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 33251 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 33252 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 33253 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.087, + "step": 33254 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 33255 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.0839, + "step": 33256 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 33257 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.0872, + "step": 33258 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 33259 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 33260 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 33261 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 33262 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 33263 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 33264 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 33265 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 33266 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 33267 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 33268 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 33269 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.0816, + "step": 33270 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 33271 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 33272 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 33273 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 33274 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 33275 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 33276 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 33277 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 33278 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.0878, + "step": 33279 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 33280 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 33281 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 33282 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 33283 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 33284 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 33285 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 33286 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 33287 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 33288 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.0875, + "step": 33289 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 33290 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 33291 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.0916, + "step": 33292 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 33293 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.085, + "step": 33294 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 33295 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 33296 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1688, + "step": 33297 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.0818, + "step": 33298 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 33299 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 33300 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 33301 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 33302 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 33303 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 33304 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.0849, + "step": 33305 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 33306 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 33307 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 33308 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.0873, + "step": 33309 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1447, + "step": 33310 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 33311 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 33312 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 33313 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 33314 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 33315 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 33316 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 33317 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 33318 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 33319 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 33320 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 33321 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 33322 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 33323 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 33324 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 33325 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 33326 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 33327 + }, + { + "epoch": 2.02, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 33328 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 33329 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 33330 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 33331 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 33332 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 33333 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 33334 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 33335 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 33336 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 33337 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 33338 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 33339 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 33340 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0952, + "step": 33341 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 33342 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 33343 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 33344 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1511, + "step": 33345 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0816, + "step": 33346 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 33347 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 33348 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 33349 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 33350 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 33351 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 33352 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 33353 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 33354 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 33355 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 33356 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 33357 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 33358 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 33359 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 33360 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 33361 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 33362 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 33363 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 33364 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 33365 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0883, + "step": 33366 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 33367 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0773, + "step": 33368 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 33369 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 33370 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 33371 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0941, + "step": 33372 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 33373 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 33374 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 33375 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 33376 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0906, + "step": 33377 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 33378 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 33379 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 33380 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 33381 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 33382 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 33383 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 33384 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 33385 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 33386 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0772, + "step": 33387 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 33388 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 33389 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 33390 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 33391 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0936, + "step": 33392 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 33393 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 33394 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 33395 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 33396 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 33397 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 33398 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 33399 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 33400 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 33401 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 33402 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 33403 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 33404 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 33405 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 33406 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 33407 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 33408 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 33409 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 33410 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 33411 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 33412 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 33413 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0863, + "step": 33414 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 33415 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 33416 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 33417 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 33418 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 33419 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1657, + "step": 33420 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 33421 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 33422 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 33423 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 33424 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 33425 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 33426 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 33427 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0771, + "step": 33428 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 33429 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 33430 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 33431 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 33432 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.145, + "step": 33433 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 33434 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 33435 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0903, + "step": 33436 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 33437 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 33438 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 33439 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 33440 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 33441 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 33442 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 33443 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 33444 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 33445 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 33446 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 33447 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 33448 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 33449 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 33450 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 33451 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0823, + "step": 33452 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 33453 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 33454 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1462, + "step": 33455 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 33456 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0819, + "step": 33457 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 33458 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 33459 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0847, + "step": 33460 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 33461 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 33462 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 33463 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 33464 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 33465 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 33466 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 33467 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 33468 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 33469 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 33470 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 33471 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 33472 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 33473 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0859, + "step": 33474 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0863, + "step": 33475 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0918, + "step": 33476 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 33477 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 33478 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 33479 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 33480 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 33481 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 33482 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 33483 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 33484 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 33485 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 33486 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 33487 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 33488 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 33489 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 33490 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0669, + "step": 33491 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 33492 + }, + { + "epoch": 2.03, + "learning_rate": 2e-05, + "loss": 0.0918, + "step": 33493 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 33494 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.0917, + "step": 33495 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 33496 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.089, + "step": 33497 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 33498 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 33499 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 33500 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 33501 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 33502 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 33503 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 33504 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1497, + "step": 33505 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 33506 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 33507 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 33508 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 33509 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 33510 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 33511 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 33512 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 33513 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 33514 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.0796, + "step": 33515 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 33516 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 33517 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 33518 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 33519 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 33520 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 33521 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 33522 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 33523 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 33524 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 33525 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 33526 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 33527 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1683, + "step": 33528 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 33529 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 33530 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 33531 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 33532 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 33533 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 33534 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 33535 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 33536 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 33537 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 33538 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 33539 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 33540 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 33541 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 33542 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 33543 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 33544 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 33545 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 33546 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 33547 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 33548 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 33549 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 33550 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 33551 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 33552 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 33553 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 33554 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 33555 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.0846, + "step": 33556 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 33557 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 33558 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 33559 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 33560 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 33561 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 33562 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 33563 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 33564 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 33565 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 33566 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 33567 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 33568 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 33569 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 33570 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.0918, + "step": 33571 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 33572 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 33573 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 33574 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 33575 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 33576 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 33577 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 33578 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 33579 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 33580 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 33581 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 33582 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 33583 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1498, + "step": 33584 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 33585 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 33586 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.0815, + "step": 33587 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 33588 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 33589 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 33590 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.0943, + "step": 33591 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 33592 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 33593 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 33594 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 33595 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 33596 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.0869, + "step": 33597 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 33598 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 33599 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 33600 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 33601 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.0845, + "step": 33602 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 33603 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 33604 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 33605 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 33606 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.0895, + "step": 33607 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 33608 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 33609 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 33610 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 33611 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 33612 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 33613 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 33614 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 33615 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 33616 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 33617 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.0839, + "step": 33618 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 33619 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 33620 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 33621 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 33622 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 33623 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 33624 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 33625 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 33626 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 33627 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 33628 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 33629 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1662, + "step": 33630 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 33631 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 33632 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 33633 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 33634 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 33635 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 33636 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 33637 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1548, + "step": 33638 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 33639 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 33640 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 33641 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 33642 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 33643 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 33644 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 33645 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 33646 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 33647 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.0824, + "step": 33648 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 33649 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 33650 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.0854, + "step": 33651 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 33652 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 33653 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.0967, + "step": 33654 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 33655 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 33656 + }, + { + "epoch": 2.04, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 33657 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 33658 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 33659 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 33660 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 33661 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 33662 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 33663 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 33664 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 33665 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 33666 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 33667 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 33668 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 33669 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.0813, + "step": 33670 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 33671 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 33672 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 33673 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 33674 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.0754, + "step": 33675 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 33676 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 33677 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 33678 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 33679 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 33680 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 33681 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 33682 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 33683 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 33684 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 33685 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 33686 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.0932, + "step": 33687 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 33688 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 33689 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 33690 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 33691 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 33692 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 33693 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 33694 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 33695 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 33696 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 33697 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 33698 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 33699 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 33700 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 33701 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 33702 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 33703 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 33704 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.0886, + "step": 33705 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 33706 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 33707 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 33708 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 33709 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 33710 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 33711 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 33712 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 33713 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.0895, + "step": 33714 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 33715 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 33716 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 33717 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 33718 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 33719 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 33720 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.09, + "step": 33721 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 33722 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 33723 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 33724 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 33725 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 33726 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 33727 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 33728 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 33729 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.0801, + "step": 33730 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 33731 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 33732 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 33733 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 33734 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 33735 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 33736 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 33737 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 33738 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 33739 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1701, + "step": 33740 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 33741 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 33742 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 33743 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 33744 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 33745 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 33746 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 33747 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 33748 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.0837, + "step": 33749 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 33750 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 33751 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 33752 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 33753 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.0808, + "step": 33754 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 33755 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 33756 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 33757 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 33758 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.084, + "step": 33759 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 33760 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 33761 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 33762 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.089, + "step": 33763 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 33764 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 33765 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 33766 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 33767 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 33768 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 33769 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 33770 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 33771 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 33772 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 33773 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 33774 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 33775 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 33776 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 33777 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 33778 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 33779 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 33780 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 33781 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.0876, + "step": 33782 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 33783 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 33784 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 33785 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 33786 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 33787 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 33788 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 33789 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 33790 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 33791 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 33792 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 33793 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 33794 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 33795 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 33796 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 33797 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 33798 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 33799 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 33800 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 33801 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 33802 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.0808, + "step": 33803 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 33804 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 33805 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 33806 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 33807 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 33808 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 33809 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 33810 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 33811 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 33812 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 33813 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.0741, + "step": 33814 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 33815 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 33816 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 33817 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 33818 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 33819 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.0796, + "step": 33820 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 33821 + }, + { + "epoch": 2.05, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 33822 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0827, + "step": 33823 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 33824 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 33825 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 33826 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 33827 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 33828 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 33829 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 33830 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 33831 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 33832 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 33833 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 33834 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 33835 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 33836 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 33837 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0747, + "step": 33838 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 33839 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 33840 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 33841 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 33842 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 33843 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 33844 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 33845 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 33846 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 33847 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 33848 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0845, + "step": 33849 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 33850 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 33851 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0856, + "step": 33852 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 33853 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 33854 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 33855 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 33856 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 33857 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 33858 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 33859 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 33860 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 33861 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 33862 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 33863 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0865, + "step": 33864 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 33865 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 33866 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 33867 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 33868 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 33869 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 33870 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 33871 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 33872 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 33873 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0871, + "step": 33874 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 33875 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 33876 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 33877 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 33878 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0915, + "step": 33879 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 33880 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 33881 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 33882 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 33883 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0881, + "step": 33884 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 33885 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 33886 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 33887 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 33888 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 33889 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 33890 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 33891 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 33892 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 33893 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 33894 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 33895 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 33896 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 33897 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 33898 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0671, + "step": 33899 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 33900 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 33901 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 33902 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 33903 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 33904 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 33905 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 33906 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 33907 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 33908 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 33909 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 33910 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 33911 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 33912 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 33913 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 33914 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1568, + "step": 33915 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 33916 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 33917 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 33918 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 33919 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 33920 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 33921 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 33922 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 33923 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 33924 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 33925 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 33926 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1591, + "step": 33927 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 33928 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 33929 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 33930 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 33931 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 33932 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 33933 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 33934 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 33935 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0952, + "step": 33936 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0822, + "step": 33937 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 33938 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 33939 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 33940 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 33941 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 33942 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 33943 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 33944 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 33945 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 33946 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 33947 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 33948 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 33949 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 33950 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0844, + "step": 33951 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0761, + "step": 33952 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 33953 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 33954 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 33955 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0824, + "step": 33956 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 33957 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 33958 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 33959 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 33960 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0932, + "step": 33961 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 33962 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 33963 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 33964 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 33965 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0891, + "step": 33966 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0855, + "step": 33967 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 33968 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0873, + "step": 33969 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0866, + "step": 33970 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 33971 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 33972 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 33973 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0929, + "step": 33974 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 33975 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 33976 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.0832, + "step": 33977 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 33978 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 33979 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 33980 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 33981 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 33982 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 33983 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 33984 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 33985 + }, + { + "epoch": 2.06, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 33986 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 33987 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 33988 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 33989 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 33990 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 33991 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 33992 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 33993 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 33994 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 33995 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 33996 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 33997 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 33998 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 33999 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 34000 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.0825, + "step": 34001 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 34002 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.0776, + "step": 34003 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1649, + "step": 34004 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 34005 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 34006 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 34007 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 34008 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 34009 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.0708, + "step": 34010 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.144, + "step": 34011 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 34012 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 34013 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 34014 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.0848, + "step": 34015 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 34016 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 34017 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 34018 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.0866, + "step": 34019 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1495, + "step": 34020 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1644, + "step": 34021 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 34022 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 34023 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 34024 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.0909, + "step": 34025 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 34026 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 34027 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 34028 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 34029 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 34030 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 34031 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 34032 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 34033 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 34034 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 34035 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 34036 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 34037 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 34038 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.0941, + "step": 34039 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 34040 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.084, + "step": 34041 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 34042 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 34043 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.0828, + "step": 34044 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 34045 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 34046 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 34047 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 34048 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.0967, + "step": 34049 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 34050 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 34051 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 34052 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.0762, + "step": 34053 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 34054 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 34055 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 34056 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 34057 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 34058 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 34059 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 34060 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 34061 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 34062 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 34063 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 34064 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 34065 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 34066 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 34067 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 34068 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.0909, + "step": 34069 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 34070 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 34071 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 34072 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 34073 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 34074 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.077, + "step": 34075 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 34076 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 34077 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 34078 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 34079 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 34080 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 34081 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.0858, + "step": 34082 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 34083 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 34084 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 34085 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 34086 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 34087 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.0777, + "step": 34088 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 34089 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 34090 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 34091 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 34092 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 34093 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 34094 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 34095 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 34096 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.0832, + "step": 34097 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 34098 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.0967, + "step": 34099 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1551, + "step": 34100 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 34101 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 34102 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 34103 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 34104 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 34105 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 34106 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 34107 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 34108 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 34109 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 34110 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 34111 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 34112 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.0633, + "step": 34113 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 34114 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 34115 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 34116 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 34117 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 34118 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 34119 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.0892, + "step": 34120 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 34121 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 34122 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 34123 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 34124 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 34125 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1488, + "step": 34126 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 34127 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 34128 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 34129 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 34130 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.0697, + "step": 34131 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 34132 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 34133 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 34134 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 34135 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 34136 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 34137 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 34138 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 34139 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 34140 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 34141 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.0832, + "step": 34142 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 34143 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 34144 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 34145 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 34146 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 34147 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 34148 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 34149 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 34150 + }, + { + "epoch": 2.07, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 34151 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.0873, + "step": 34152 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 34153 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 34154 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 34155 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 34156 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 34157 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 34158 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 34159 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 34160 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 34161 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 34162 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 34163 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 34164 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.0767, + "step": 34165 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 34166 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 34167 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 34168 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 34169 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 34170 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 34171 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.0898, + "step": 34172 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.0842, + "step": 34173 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 34174 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 34175 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 34176 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 34177 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 34178 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1503, + "step": 34179 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 34180 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 34181 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 34182 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 34183 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 34184 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 34185 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 34186 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.0806, + "step": 34187 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 34188 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1579, + "step": 34189 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 34190 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 34191 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 34192 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 34193 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 34194 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 34195 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 34196 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 34197 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 34198 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 34199 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 34200 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 34201 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 34202 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 34203 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 34204 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 34205 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 34206 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 34207 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 34208 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 34209 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 34210 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 34211 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1505, + "step": 34212 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 34213 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 34214 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 34215 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 34216 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 34217 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 34218 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 34219 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 34220 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1736, + "step": 34221 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 34222 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 34223 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 34224 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 34225 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 34226 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 34227 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 34228 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 34229 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 34230 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 34231 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 34232 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 34233 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.0874, + "step": 34234 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 34235 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 34236 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 34237 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 34238 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 34239 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 34240 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 34241 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 34242 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 34243 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 34244 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 34245 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 34246 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1558, + "step": 34247 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 34248 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.09, + "step": 34249 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 34250 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 34251 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1495, + "step": 34252 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 34253 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 34254 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 34255 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 34256 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.09, + "step": 34257 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 34258 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 34259 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 34260 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 34261 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.087, + "step": 34262 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 34263 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 34264 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 34265 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1568, + "step": 34266 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1654, + "step": 34267 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1666, + "step": 34268 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 34269 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 34270 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 34271 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 34272 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 34273 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 34274 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 34275 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 34276 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 34277 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.148, + "step": 34278 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 34279 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.0889, + "step": 34280 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 34281 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 34282 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 34283 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 34284 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 34285 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 34286 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 34287 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 34288 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 34289 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 34290 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 34291 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 34292 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 34293 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 34294 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 34295 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 34296 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 34297 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 34298 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 34299 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 34300 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.087, + "step": 34301 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 34302 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 34303 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 34304 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 34305 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 34306 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 34307 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.0776, + "step": 34308 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 34309 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 34310 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.0914, + "step": 34311 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 34312 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 34313 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 34314 + }, + { + "epoch": 2.08, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 34315 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 34316 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 34317 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 34318 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 34319 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 34320 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 34321 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.0871, + "step": 34322 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 34323 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 34324 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 34325 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 34326 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 34327 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 34328 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 34329 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 34330 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 34331 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 34332 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 34333 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 34334 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1814, + "step": 34335 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 34336 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 34337 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 34338 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 34339 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 34340 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.0918, + "step": 34341 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 34342 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 34343 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 34344 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 34345 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 34346 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 34347 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 34348 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 34349 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 34350 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.09, + "step": 34351 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 34352 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 34353 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 34354 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 34355 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 34356 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 34357 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.0967, + "step": 34358 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 34359 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 34360 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 34361 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 34362 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 34363 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.0727, + "step": 34364 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.0846, + "step": 34365 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 34366 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 34367 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 34368 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 34369 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 34370 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.084, + "step": 34371 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 34372 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.0902, + "step": 34373 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.0825, + "step": 34374 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 34375 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 34376 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 34377 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 34378 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.0877, + "step": 34379 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 34380 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 34381 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 34382 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 34383 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 34384 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 34385 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.0894, + "step": 34386 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.0871, + "step": 34387 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 34388 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 34389 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 34390 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 34391 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 34392 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 34393 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 34394 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 34395 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 34396 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 34397 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 34398 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1849, + "step": 34399 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 34400 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 34401 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 34402 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 34403 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 34404 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1578, + "step": 34405 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 34406 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 34407 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 34408 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 34409 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 34410 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 34411 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 34412 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 34413 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 34414 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 34415 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 34416 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 34417 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 34418 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 34419 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 34420 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 34421 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 34422 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 34423 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 34424 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 34425 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 34426 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.147, + "step": 34427 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 34428 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 34429 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 34430 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 34431 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 34432 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 34433 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.0836, + "step": 34434 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.0915, + "step": 34435 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 34436 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 34437 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.0859, + "step": 34438 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 34439 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 34440 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 34441 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.0802, + "step": 34442 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 34443 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 34444 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 34445 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 34446 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 34447 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.0932, + "step": 34448 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 34449 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 34450 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.0917, + "step": 34451 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 34452 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1517, + "step": 34453 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 34454 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 34455 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 34456 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 34457 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 34458 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 34459 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 34460 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 34461 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.0902, + "step": 34462 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 34463 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 34464 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 34465 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 34466 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 34467 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 34468 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.0842, + "step": 34469 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 34470 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.0778, + "step": 34471 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 34472 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 34473 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 34474 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 34475 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.0917, + "step": 34476 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 34477 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1558, + "step": 34478 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 34479 + }, + { + "epoch": 2.09, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 34480 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 34481 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 34482 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 34483 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 34484 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 34485 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 34486 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 34487 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 34488 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 34489 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 34490 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 34491 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 34492 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 34493 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 34494 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 34495 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 34496 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 34497 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 34498 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 34499 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 34500 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 34501 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 34502 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 34503 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 34504 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 34505 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 34506 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 34507 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 34508 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 34509 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 34510 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 34511 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 34512 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 34513 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 34514 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 34515 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 34516 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 34517 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 34518 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.0878, + "step": 34519 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1646, + "step": 34520 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 34521 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 34522 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 34523 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.0769, + "step": 34524 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 34525 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 34526 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 34527 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.0886, + "step": 34528 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.0932, + "step": 34529 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 34530 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 34531 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 34532 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 34533 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 34534 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.0785, + "step": 34535 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 34536 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 34537 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 34538 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 34539 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 34540 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 34541 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 34542 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 34543 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 34544 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 34545 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 34546 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 34547 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 34548 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 34549 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1453, + "step": 34550 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 34551 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 34552 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 34553 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.0889, + "step": 34554 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 34555 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 34556 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 34557 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 34558 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 34559 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 34560 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 34561 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 34562 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 34563 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 34564 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 34565 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 34566 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 34567 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 34568 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 34569 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 34570 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.0864, + "step": 34571 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 34572 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 34573 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 34574 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 34575 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 34576 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 34577 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 34578 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 34579 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 34580 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 34581 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 34582 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 34583 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 34584 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 34585 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 34586 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 34587 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 34588 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 34589 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 34590 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 34591 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 34592 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 34593 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 34594 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1451, + "step": 34595 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 34596 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.0789, + "step": 34597 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 34598 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 34599 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 34600 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 34601 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 34602 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 34603 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 34604 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 34605 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 34606 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.0888, + "step": 34607 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 34608 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 34609 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 34610 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 34611 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 34612 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 34613 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 34614 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 34615 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 34616 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 34617 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 34618 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 34619 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 34620 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.0936, + "step": 34621 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 34622 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 34623 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 34624 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 34625 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 34626 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 34627 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 34628 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 34629 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 34630 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 34631 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 34632 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 34633 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 34634 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 34635 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 34636 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 34637 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 34638 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 34639 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 34640 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 34641 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.0886, + "step": 34642 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 34643 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 34644 + }, + { + "epoch": 2.1, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 34645 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 34646 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 34647 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 34648 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 34649 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 34650 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 34651 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 34652 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 34653 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 34654 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.0877, + "step": 34655 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 34656 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.0887, + "step": 34657 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 34658 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 34659 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 34660 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.081, + "step": 34661 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 34662 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 34663 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 34664 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 34665 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 34666 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 34667 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 34668 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 34669 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 34670 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 34671 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 34672 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 34673 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 34674 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 34675 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 34676 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 34677 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 34678 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 34679 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 34680 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 34681 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 34682 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 34683 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.0914, + "step": 34684 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 34685 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1462, + "step": 34686 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 34687 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 34688 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 34689 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 34690 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 34691 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 34692 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1528, + "step": 34693 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 34694 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 34695 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 34696 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 34697 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 34698 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 34699 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 34700 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 34701 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.0902, + "step": 34702 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.0864, + "step": 34703 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 34704 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 34705 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 34706 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 34707 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 34708 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.0865, + "step": 34709 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 34710 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 34711 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.0771, + "step": 34712 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 34713 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 34714 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 34715 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 34716 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 34717 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 34718 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 34719 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1568, + "step": 34720 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 34721 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 34722 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 34723 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.0741, + "step": 34724 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 34725 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 34726 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 34727 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 34728 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 34729 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 34730 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 34731 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 34732 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 34733 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 34734 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 34735 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 34736 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 34737 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 34738 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 34739 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 34740 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 34741 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 34742 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 34743 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 34744 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 34745 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 34746 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 34747 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 34748 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 34749 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 34750 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 34751 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 34752 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 34753 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 34754 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.0767, + "step": 34755 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 34756 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 34757 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 34758 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 34759 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 34760 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 34761 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 34762 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1631, + "step": 34763 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 34764 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 34765 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 34766 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 34767 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.0888, + "step": 34768 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1722, + "step": 34769 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 34770 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 34771 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 34772 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 34773 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 34774 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.0883, + "step": 34775 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 34776 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.0916, + "step": 34777 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 34778 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 34779 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1583, + "step": 34780 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1443, + "step": 34781 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 34782 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 34783 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 34784 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 34785 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 34786 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 34787 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 34788 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.0806, + "step": 34789 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 34790 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.089, + "step": 34791 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 34792 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 34793 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 34794 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 34795 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 34796 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 34797 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 34798 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 34799 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 34800 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 34801 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 34802 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 34803 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 34804 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 34805 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 34806 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 34807 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 34808 + }, + { + "epoch": 2.11, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 34809 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.0787, + "step": 34810 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 34811 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1443, + "step": 34812 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 34813 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.082, + "step": 34814 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 34815 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.079, + "step": 34816 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 34817 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 34818 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 34819 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 34820 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 34821 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 34822 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 34823 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 34824 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.0876, + "step": 34825 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 34826 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 34827 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 34828 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 34829 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 34830 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 34831 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 34832 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 34833 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 34834 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 34835 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 34836 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 34837 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 34838 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 34839 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 34840 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 34841 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 34842 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1557, + "step": 34843 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 34844 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 34845 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.0805, + "step": 34846 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 34847 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 34848 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 34849 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 34850 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 34851 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 34852 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 34853 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.0815, + "step": 34854 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 34855 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 34856 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 34857 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 34858 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 34859 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 34860 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 34861 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 34862 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 34863 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 34864 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 34865 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.0876, + "step": 34866 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.0952, + "step": 34867 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 34868 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.0855, + "step": 34869 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 34870 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.081, + "step": 34871 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 34872 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 34873 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 34874 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 34875 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 34876 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 34877 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.0776, + "step": 34878 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 34879 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 34880 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 34881 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 34882 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 34883 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.0826, + "step": 34884 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 34885 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 34886 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1904, + "step": 34887 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 34888 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1645, + "step": 34889 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.0828, + "step": 34890 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 34891 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 34892 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.0864, + "step": 34893 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 34894 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 34895 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 34896 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 34897 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 34898 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 34899 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 34900 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 34901 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 34902 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 34903 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 34904 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 34905 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.087, + "step": 34906 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 34907 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 34908 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 34909 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 34910 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 34911 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 34912 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 34913 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 34914 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 34915 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 34916 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 34917 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 34918 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 34919 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 34920 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 34921 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 34922 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 34923 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 34924 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 34925 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 34926 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 34927 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 34928 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 34929 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 34930 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 34931 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 34932 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 34933 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 34934 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 34935 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 34936 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 34937 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 34938 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 34939 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 34940 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 34941 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 34942 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 34943 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 34944 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 34945 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 34946 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 34947 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 34948 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 34949 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 34950 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.0817, + "step": 34951 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.0891, + "step": 34952 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 34953 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 34954 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 34955 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 34956 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 34957 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 34958 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 34959 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 34960 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 34961 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 34962 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 34963 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 34964 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 34965 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 34966 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 34967 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 34968 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 34969 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 34970 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 34971 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 34972 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.0932, + "step": 34973 + }, + { + "epoch": 2.12, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 34974 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 34975 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 34976 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 34977 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 34978 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 34979 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 34980 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.0871, + "step": 34981 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 34982 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 34983 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 34984 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 34985 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.0839, + "step": 34986 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 34987 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 34988 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 34989 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 34990 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 34991 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 34992 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 34993 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 34994 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 34995 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 34996 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 34997 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 34998 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.0891, + "step": 34999 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 35000 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 35001 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 35002 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 35003 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 35004 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 35005 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 35006 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 35007 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 35008 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 35009 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 35010 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 35011 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 35012 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 35013 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1553, + "step": 35014 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 35015 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.0891, + "step": 35016 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 35017 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 35018 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 35019 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 35020 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 35021 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 35022 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 35023 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 35024 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.0903, + "step": 35025 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 35026 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 35027 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 35028 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 35029 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 35030 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 35031 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 35032 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 35033 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 35034 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 35035 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 35036 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 35037 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 35038 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 35039 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 35040 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.0903, + "step": 35041 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 35042 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 35043 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 35044 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 35045 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 35046 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 35047 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 35048 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 35049 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 35050 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 35051 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 35052 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 35053 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 35054 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 35055 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 35056 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 35057 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 35058 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 35059 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 35060 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 35061 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 35062 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 35063 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 35064 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 35065 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 35066 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 35067 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 35068 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 35069 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 35070 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.0882, + "step": 35071 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 35072 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 35073 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 35074 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 35075 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.0807, + "step": 35076 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 35077 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.0881, + "step": 35078 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 35079 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 35080 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1614, + "step": 35081 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 35082 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 35083 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 35084 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 35085 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 35086 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 35087 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 35088 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 35089 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 35090 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 35091 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 35092 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 35093 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 35094 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 35095 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 35096 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 35097 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 35098 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.09, + "step": 35099 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 35100 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 35101 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 35102 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 35103 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 35104 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.0915, + "step": 35105 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 35106 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.0888, + "step": 35107 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.0692, + "step": 35108 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 35109 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1517, + "step": 35110 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 35111 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 35112 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 35113 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 35114 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 35115 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.0875, + "step": 35116 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 35117 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 35118 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 35119 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 35120 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 35121 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.0863, + "step": 35122 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 35123 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 35124 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 35125 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 35126 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 35127 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 35128 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 35129 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 35130 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 35131 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 35132 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 35133 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 35134 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 35135 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 35136 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 35137 + }, + { + "epoch": 2.13, + "learning_rate": 2e-05, + "loss": 0.082, + "step": 35138 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 35139 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 35140 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 35141 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 35142 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 35143 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 35144 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 35145 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 35146 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 35147 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 35148 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 35149 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 35150 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.0854, + "step": 35151 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 35152 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 35153 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 35154 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 35155 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.0895, + "step": 35156 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 35157 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 35158 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 35159 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.0915, + "step": 35160 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 35161 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 35162 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 35163 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 35164 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 35165 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 35166 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 35167 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 35168 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 35169 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 35170 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 35171 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 35172 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 35173 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 35174 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 35175 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 35176 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 35177 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 35178 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 35179 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 35180 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.071, + "step": 35181 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 35182 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 35183 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 35184 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.0894, + "step": 35185 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.0864, + "step": 35186 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 35187 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 35188 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 35189 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.0773, + "step": 35190 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 35191 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 35192 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1456, + "step": 35193 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 35194 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 35195 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 35196 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 35197 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1728, + "step": 35198 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 35199 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 35200 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 35201 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 35202 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 35203 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 35204 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 35205 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 35206 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 35207 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 35208 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 35209 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 35210 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 35211 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1513, + "step": 35212 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 35213 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 35214 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 35215 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 35216 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 35217 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 35218 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 35219 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1534, + "step": 35220 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 35221 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 35222 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 35223 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 35224 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.0914, + "step": 35225 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 35226 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 35227 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 35228 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1579, + "step": 35229 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 35230 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 35231 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1488, + "step": 35232 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 35233 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.0881, + "step": 35234 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 35235 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 35236 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 35237 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 35238 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 35239 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 35240 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 35241 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 35242 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 35243 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 35244 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.0803, + "step": 35245 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1683, + "step": 35246 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 35247 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 35248 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 35249 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 35250 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 35251 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 35252 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 35253 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.0897, + "step": 35254 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 35255 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 35256 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 35257 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 35258 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 35259 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 35260 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 35261 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.0898, + "step": 35262 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 35263 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 35264 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 35265 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 35266 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 35267 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 35268 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 35269 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 35270 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.152, + "step": 35271 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 35272 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1511, + "step": 35273 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 35274 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.0872, + "step": 35275 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 35276 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 35277 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 35278 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 35279 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 35280 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.0838, + "step": 35281 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.0816, + "step": 35282 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 35283 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1478, + "step": 35284 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 35285 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 35286 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 35287 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 35288 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.0936, + "step": 35289 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 35290 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 35291 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1793, + "step": 35292 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 35293 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 35294 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 35295 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 35296 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 35297 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 35298 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 35299 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 35300 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 35301 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 35302 + }, + { + "epoch": 2.14, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 35303 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 35304 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 35305 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 35306 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 35307 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 35308 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 35309 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 35310 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 35311 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 35312 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.0878, + "step": 35313 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 35314 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 35315 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 35316 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 35317 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.0854, + "step": 35318 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 35319 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 35320 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 35321 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 35322 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 35323 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 35324 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 35325 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 35326 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 35327 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.0846, + "step": 35328 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 35329 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 35330 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 35331 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 35332 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 35333 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 35334 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 35335 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 35336 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 35337 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 35338 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1805, + "step": 35339 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 35340 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 35341 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 35342 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 35343 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 35344 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 35345 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 35346 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 35347 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 35348 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 35349 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 35350 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 35351 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 35352 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 35353 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 35354 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 35355 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 35356 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 35357 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 35358 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 35359 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 35360 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 35361 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 35362 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 35363 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 35364 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 35365 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 35366 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 35367 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 35368 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 35369 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 35370 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 35371 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 35372 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 35373 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 35374 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 35375 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 35376 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1502, + "step": 35377 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 35378 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.0902, + "step": 35379 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 35380 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 35381 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1746, + "step": 35382 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 35383 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 35384 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 35385 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.0941, + "step": 35386 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 35387 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 35388 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 35389 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 35390 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 35391 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 35392 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 35393 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 35394 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 35395 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 35396 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 35397 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 35398 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 35399 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 35400 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 35401 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 35402 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 35403 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 35404 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 35405 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 35406 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 35407 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 35408 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 35409 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 35410 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 35411 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 35412 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 35413 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 35414 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 35415 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 35416 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.0817, + "step": 35417 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 35418 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 35419 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 35420 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 35421 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.0848, + "step": 35422 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 35423 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.0846, + "step": 35424 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.0828, + "step": 35425 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 35426 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 35427 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 35428 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 35429 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 35430 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 35431 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 35432 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 35433 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 35434 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1608, + "step": 35435 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 35436 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 35437 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 35438 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 35439 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 35440 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 35441 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 35442 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 35443 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1791, + "step": 35444 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 35445 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 35446 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 35447 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.0824, + "step": 35448 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 35449 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 35450 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 35451 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 35452 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 35453 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 35454 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 35455 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1662, + "step": 35456 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 35457 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 35458 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 35459 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 35460 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 35461 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 35462 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 35463 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 35464 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 35465 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 35466 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.0882, + "step": 35467 + }, + { + "epoch": 2.15, + "learning_rate": 2e-05, + "loss": 0.0764, + "step": 35468 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 35469 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 35470 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 35471 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 35472 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.0887, + "step": 35473 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 35474 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 35475 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 35476 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.0869, + "step": 35477 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.0935, + "step": 35478 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 35479 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 35480 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 35481 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 35482 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 35483 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 35484 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 35485 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.0911, + "step": 35486 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 35487 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 35488 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 35489 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 35490 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 35491 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 35492 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 35493 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.0881, + "step": 35494 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 35495 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 35496 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1631, + "step": 35497 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.0897, + "step": 35498 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 35499 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 35500 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 35501 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 35502 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 35503 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 35504 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 35505 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 35506 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 35507 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.0897, + "step": 35508 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1579, + "step": 35509 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 35510 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 35511 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 35512 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 35513 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.0852, + "step": 35514 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 35515 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 35516 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 35517 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 35518 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.0836, + "step": 35519 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 35520 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 35521 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 35522 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 35523 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 35524 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 35525 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 35526 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 35527 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 35528 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 35529 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 35530 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 35531 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 35532 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 35533 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 35534 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 35535 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 35536 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 35537 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 35538 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 35539 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 35540 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 35541 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 35542 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 35543 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 35544 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.0898, + "step": 35545 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 35546 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 35547 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 35548 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 35549 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 35550 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.0915, + "step": 35551 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 35552 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 35553 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.0757, + "step": 35554 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.0897, + "step": 35555 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.0675, + "step": 35556 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 35557 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.0733, + "step": 35558 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 35559 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 35560 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 35561 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 35562 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 35563 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 35564 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 35565 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 35566 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 35567 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 35568 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.0785, + "step": 35569 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 35570 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 35571 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 35572 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 35573 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 35574 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 35575 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 35576 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 35577 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 35578 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 35579 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 35580 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 35581 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.0863, + "step": 35582 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 35583 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 35584 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 35585 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 35586 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 35587 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 35588 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 35589 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.174, + "step": 35590 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 35591 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 35592 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 35593 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 35594 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 35595 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 35596 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 35597 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 35598 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 35599 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 35600 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 35601 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 35602 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 35603 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 35604 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 35605 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 35606 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 35607 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.0847, + "step": 35608 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 35609 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.0853, + "step": 35610 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 35611 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 35612 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 35613 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 35614 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 35615 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 35616 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 35617 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 35618 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1541, + "step": 35619 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 35620 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 35621 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 35622 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 35623 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 35624 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 35625 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 35626 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 35627 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 35628 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 35629 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 35630 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.0817, + "step": 35631 + }, + { + "epoch": 2.16, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 35632 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 35633 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.0792, + "step": 35634 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 35635 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 35636 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 35637 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 35638 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 35639 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 35640 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 35641 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 35642 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 35643 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.0892, + "step": 35644 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 35645 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 35646 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 35647 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 35648 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 35649 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 35650 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 35651 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 35652 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.0793, + "step": 35653 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.0876, + "step": 35654 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 35655 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 35656 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 35657 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 35658 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 35659 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 35660 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 35661 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 35662 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 35663 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 35664 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 35665 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 35666 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.0791, + "step": 35667 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 35668 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 35669 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 35670 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.0822, + "step": 35671 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 35672 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 35673 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 35674 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 35675 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 35676 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 35677 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 35678 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.0967, + "step": 35679 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 35680 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 35681 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 35682 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 35683 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 35684 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 35685 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 35686 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 35687 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.0825, + "step": 35688 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 35689 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 35690 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 35691 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 35692 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.0721, + "step": 35693 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 35694 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 35695 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.0868, + "step": 35696 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 35697 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 35698 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 35699 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 35700 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 35701 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 35702 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 35703 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 35704 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.0868, + "step": 35705 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 35706 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 35707 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 35708 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 35709 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 35710 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 35711 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 35712 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 35713 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 35714 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.0909, + "step": 35715 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 35716 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 35717 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 35718 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 35719 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 35720 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 35721 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 35722 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 35723 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 35724 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 35725 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.16, + "step": 35726 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 35727 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 35728 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 35729 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 35730 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 35731 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.0867, + "step": 35732 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 35733 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 35734 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 35735 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 35736 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.154, + "step": 35737 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 35738 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 35739 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 35740 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 35741 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 35742 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.0967, + "step": 35743 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 35744 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 35745 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 35746 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.0877, + "step": 35747 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 35748 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 35749 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.0851, + "step": 35750 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 35751 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 35752 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 35753 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 35754 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.0877, + "step": 35755 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.091, + "step": 35756 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 35757 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 35758 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 35759 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.0874, + "step": 35760 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 35761 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 35762 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 35763 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 35764 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1469, + "step": 35765 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 35766 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 35767 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 35768 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 35769 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 35770 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.0889, + "step": 35771 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 35772 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 35773 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 35774 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 35775 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 35776 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 35777 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 35778 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 35779 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 35780 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 35781 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 35782 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 35783 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 35784 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 35785 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 35786 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 35787 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 35788 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 35789 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 35790 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 35791 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 35792 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1443, + "step": 35793 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 35794 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 35795 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 35796 + }, + { + "epoch": 2.17, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 35797 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 35798 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 35799 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.0918, + "step": 35800 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 35801 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 35802 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1597, + "step": 35803 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.0911, + "step": 35804 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 35805 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 35806 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.077, + "step": 35807 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 35808 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 35809 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 35810 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 35811 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 35812 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 35813 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 35814 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 35815 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 35816 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 35817 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 35818 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 35819 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 35820 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 35821 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 35822 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 35823 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 35824 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1669, + "step": 35825 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 35826 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 35827 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 35828 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 35829 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 35830 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 35831 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 35832 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 35833 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 35834 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 35835 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 35836 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 35837 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 35838 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 35839 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 35840 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 35841 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 35842 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 35843 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 35844 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 35845 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 35846 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 35847 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 35848 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 35849 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 35850 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 35851 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 35852 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 35853 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 35854 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 35855 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 35856 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 35857 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 35858 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 35859 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 35860 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 35861 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 35862 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 35863 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.0804, + "step": 35864 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 35865 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 35866 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 35867 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 35868 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 35869 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 35870 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 35871 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 35872 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 35873 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 35874 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 35875 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 35876 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 35877 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 35878 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 35879 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 35880 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 35881 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 35882 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 35883 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1549, + "step": 35884 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 35885 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 35886 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 35887 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 35888 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 35889 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 35890 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 35891 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 35892 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 35893 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 35894 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 35895 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 35896 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 35897 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.0828, + "step": 35898 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 35899 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 35900 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 35901 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 35902 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 35903 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 35904 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 35905 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.0852, + "step": 35906 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 35907 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 35908 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 35909 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 35910 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 35911 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 35912 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 35913 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 35914 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 35915 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 35916 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 35917 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 35918 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 35919 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 35920 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 35921 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 35922 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 35923 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 35924 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 35925 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 35926 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 35927 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.0917, + "step": 35928 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 35929 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 35930 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 35931 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 35932 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 35933 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 35934 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 35935 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 35936 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 35937 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 35938 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 35939 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 35940 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 35941 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 35942 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 35943 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.0867, + "step": 35944 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 35945 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 35946 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 35947 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 35948 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 35949 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 35950 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.0898, + "step": 35951 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 35952 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 35953 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 35954 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 35955 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 35956 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 35957 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 35958 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 35959 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1519, + "step": 35960 + }, + { + "epoch": 2.18, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 35961 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 35962 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 35963 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 35964 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 35965 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 35966 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 35967 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 35968 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 35969 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 35970 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 35971 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 35972 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 35973 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 35974 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 35975 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 35976 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 35977 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 35978 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 35979 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 35980 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 35981 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 35982 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 35983 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 35984 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 35985 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 35986 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 35987 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 35988 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 35989 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 35990 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 35991 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.0873, + "step": 35992 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.0911, + "step": 35993 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.0873, + "step": 35994 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 35995 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 35996 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 35997 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 35998 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 35999 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 36000 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 36001 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 36002 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 36003 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 36004 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 36005 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 36006 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 36007 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 36008 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 36009 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 36010 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 36011 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.0952, + "step": 36012 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 36013 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 36014 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 36015 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 36016 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 36017 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 36018 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 36019 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 36020 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 36021 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 36022 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 36023 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 36024 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1488, + "step": 36025 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 36026 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 36027 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 36028 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 36029 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.0836, + "step": 36030 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 36031 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 36032 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 36033 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 36034 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.0915, + "step": 36035 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 36036 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 36037 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 36038 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1469, + "step": 36039 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 36040 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 36041 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 36042 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 36043 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 36044 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 36045 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 36046 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.0866, + "step": 36047 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 36048 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 36049 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 36050 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 36051 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 36052 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 36053 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 36054 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 36055 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 36056 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 36057 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 36058 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 36059 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.0826, + "step": 36060 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 36061 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 36062 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.0943, + "step": 36063 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.0941, + "step": 36064 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.0814, + "step": 36065 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 36066 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 36067 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 36068 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 36069 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 36070 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 36071 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 36072 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 36073 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 36074 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.0721, + "step": 36075 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 36076 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 36077 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 36078 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.0882, + "step": 36079 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 36080 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 36081 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 36082 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 36083 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 36084 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.0859, + "step": 36085 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.0824, + "step": 36086 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 36087 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 36088 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 36089 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 36090 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 36091 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 36092 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 36093 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 36094 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 36095 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.0867, + "step": 36096 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 36097 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 36098 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 36099 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 36100 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 36101 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 36102 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 36103 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.0952, + "step": 36104 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 36105 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 36106 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 36107 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 36108 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.0782, + "step": 36109 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 36110 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 36111 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 36112 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 36113 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 36114 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 36115 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 36116 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 36117 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 36118 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.0917, + "step": 36119 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.0789, + "step": 36120 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 36121 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 36122 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 36123 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 36124 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 36125 + }, + { + "epoch": 2.19, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 36126 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 36127 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 36128 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.0892, + "step": 36129 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 36130 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 36131 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 36132 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 36133 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 36134 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 36135 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 36136 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 36137 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 36138 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.0817, + "step": 36139 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.0952, + "step": 36140 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 36141 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1645, + "step": 36142 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 36143 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 36144 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 36145 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 36146 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 36147 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 36148 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 36149 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1586, + "step": 36150 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 36151 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 36152 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 36153 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 36154 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 36155 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 36156 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 36157 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 36158 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 36159 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 36160 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 36161 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1453, + "step": 36162 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 36163 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 36164 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 36165 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.0806, + "step": 36166 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 36167 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.0791, + "step": 36168 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 36169 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 36170 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 36171 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.0872, + "step": 36172 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 36173 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 36174 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 36175 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 36176 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 36177 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 36178 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 36179 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 36180 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 36181 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 36182 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 36183 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.081, + "step": 36184 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 36185 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 36186 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 36187 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 36188 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 36189 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 36190 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 36191 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 36192 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.078, + "step": 36193 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 36194 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 36195 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 36196 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 36197 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 36198 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 36199 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 36200 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 36201 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 36202 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.0751, + "step": 36203 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 36204 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 36205 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 36206 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 36207 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 36208 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 36209 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 36210 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 36211 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 36212 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1633, + "step": 36213 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 36214 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 36215 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 36216 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 36217 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 36218 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.0814, + "step": 36219 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 36220 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 36221 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 36222 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 36223 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 36224 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.0901, + "step": 36225 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 36226 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 36227 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 36228 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 36229 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 36230 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 36231 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 36232 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 36233 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 36234 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 36235 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 36236 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 36237 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 36238 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.0892, + "step": 36239 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 36240 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 36241 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 36242 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1552, + "step": 36243 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.0806, + "step": 36244 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 36245 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 36246 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 36247 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 36248 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 36249 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 36250 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 36251 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.147, + "step": 36252 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 36253 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 36254 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 36255 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 36256 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 36257 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1526, + "step": 36258 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 36259 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 36260 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 36261 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 36262 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.0915, + "step": 36263 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 36264 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 36265 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 36266 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.0811, + "step": 36267 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 36268 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.0901, + "step": 36269 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 36270 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 36271 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 36272 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 36273 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 36274 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 36275 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 36276 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1526, + "step": 36277 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 36278 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 36279 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 36280 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 36281 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 36282 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 36283 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 36284 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 36285 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.0903, + "step": 36286 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 36287 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 36288 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 36289 + }, + { + "epoch": 2.2, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 36290 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 36291 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 36292 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 36293 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 36294 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.0851, + "step": 36295 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 36296 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 36297 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 36298 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 36299 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 36300 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 36301 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.0853, + "step": 36302 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 36303 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 36304 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 36305 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 36306 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 36307 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 36308 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 36309 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 36310 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 36311 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 36312 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 36313 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 36314 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 36315 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 36316 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 36317 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 36318 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 36319 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 36320 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 36321 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 36322 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 36323 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 36324 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 36325 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 36326 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 36327 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 36328 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 36329 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1606, + "step": 36330 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 36331 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 36332 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 36333 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 36334 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 36335 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 36336 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 36337 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 36338 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 36339 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 36340 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.0693, + "step": 36341 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 36342 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 36343 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 36344 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 36345 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 36346 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1499, + "step": 36347 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 36348 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 36349 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 36350 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 36351 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 36352 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 36353 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 36354 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 36355 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 36356 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 36357 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 36358 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.0811, + "step": 36359 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 36360 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 36361 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 36362 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 36363 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 36364 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.0882, + "step": 36365 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 36366 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 36367 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 36368 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 36369 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 36370 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.0902, + "step": 36371 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 36372 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 36373 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 36374 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 36375 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 36376 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 36377 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 36378 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 36379 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 36380 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 36381 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 36382 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 36383 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 36384 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 36385 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 36386 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 36387 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 36388 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 36389 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.0868, + "step": 36390 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 36391 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 36392 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 36393 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 36394 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 36395 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 36396 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.0901, + "step": 36397 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 36398 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 36399 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 36400 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 36401 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 36402 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 36403 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 36404 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 36405 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 36406 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 36407 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 36408 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 36409 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 36410 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 36411 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 36412 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 36413 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 36414 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 36415 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1524, + "step": 36416 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 36417 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 36418 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1511, + "step": 36419 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 36420 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 36421 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 36422 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 36423 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 36424 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 36425 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 36426 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 36427 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 36428 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 36429 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 36430 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 36431 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 36432 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 36433 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 36434 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 36435 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 36436 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 36437 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 36438 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 36439 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 36440 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1566, + "step": 36441 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 36442 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 36443 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 36444 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 36445 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 36446 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 36447 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 36448 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 36449 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 36450 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 36451 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 36452 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 36453 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1515, + "step": 36454 + }, + { + "epoch": 2.21, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 36455 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 36456 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 36457 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 36458 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 36459 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 36460 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.0889, + "step": 36461 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 36462 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 36463 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 36464 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 36465 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 36466 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 36467 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1505, + "step": 36468 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 36469 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.0883, + "step": 36470 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 36471 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 36472 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 36473 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 36474 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 36475 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 36476 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 36477 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 36478 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 36479 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.0894, + "step": 36480 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 36481 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 36482 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 36483 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 36484 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1749, + "step": 36485 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 36486 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 36487 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 36488 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 36489 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 36490 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 36491 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 36492 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1592, + "step": 36493 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1588, + "step": 36494 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 36495 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 36496 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 36497 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 36498 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 36499 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 36500 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 36501 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 36502 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 36503 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 36504 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 36505 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 36506 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.0936, + "step": 36507 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 36508 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 36509 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 36510 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 36511 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 36512 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 36513 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 36514 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 36515 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 36516 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.088, + "step": 36517 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 36518 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 36519 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 36520 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 36521 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 36522 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 36523 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 36524 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 36525 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 36526 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 36527 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 36528 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 36529 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 36530 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 36531 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 36532 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.0799, + "step": 36533 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 36534 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 36535 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.088, + "step": 36536 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 36537 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 36538 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 36539 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 36540 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 36541 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 36542 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 36543 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.0819, + "step": 36544 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 36545 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 36546 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.0864, + "step": 36547 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 36548 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 36549 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 36550 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 36551 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 36552 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 36553 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 36554 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.155, + "step": 36555 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.0849, + "step": 36556 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 36557 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 36558 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 36559 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.089, + "step": 36560 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 36561 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 36562 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 36563 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 36564 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 36565 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 36566 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 36567 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 36568 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 36569 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.0843, + "step": 36570 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 36571 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 36572 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1447, + "step": 36573 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 36574 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 36575 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 36576 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 36577 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 36578 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 36579 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 36580 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 36581 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 36582 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 36583 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 36584 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 36585 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 36586 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 36587 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 36588 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 36589 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 36590 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.0898, + "step": 36591 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 36592 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 36593 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 36594 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 36595 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 36596 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 36597 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 36598 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 36599 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 36600 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 36601 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 36602 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 36603 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 36604 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 36605 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.0901, + "step": 36606 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 36607 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 36608 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 36609 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 36610 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.0916, + "step": 36611 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 36612 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 36613 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 36614 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 36615 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 36616 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 36617 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 36618 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 36619 + }, + { + "epoch": 2.22, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 36620 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 36621 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 36622 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.0858, + "step": 36623 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 36624 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 36625 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 36626 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 36627 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.0816, + "step": 36628 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 36629 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1588, + "step": 36630 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 36631 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 36632 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 36633 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 36634 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 36635 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 36636 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 36637 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 36638 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 36639 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 36640 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 36641 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 36642 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 36643 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 36644 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 36645 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 36646 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 36647 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 36648 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 36649 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 36650 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 36651 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 36652 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 36653 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 36654 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 36655 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 36656 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 36657 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 36658 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 36659 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 36660 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 36661 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 36662 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 36663 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1686, + "step": 36664 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 36665 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 36666 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 36667 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.0864, + "step": 36668 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 36669 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.0887, + "step": 36670 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 36671 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 36672 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 36673 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 36674 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 36675 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.0929, + "step": 36676 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 36677 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 36678 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 36679 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 36680 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 36681 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 36682 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 36683 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 36684 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 36685 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 36686 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 36687 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 36688 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 36689 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.0773, + "step": 36690 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 36691 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 36692 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 36693 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 36694 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 36695 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 36696 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 36697 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 36698 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 36699 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 36700 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 36701 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 36702 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 36703 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 36704 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 36705 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 36706 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 36707 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 36708 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 36709 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1429, + "step": 36710 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.0888, + "step": 36711 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 36712 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.0794, + "step": 36713 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 36714 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 36715 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 36716 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 36717 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 36718 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 36719 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 36720 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.0935, + "step": 36721 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 36722 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 36723 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 36724 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.0851, + "step": 36725 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 36726 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 36727 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 36728 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 36729 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 36730 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.0864, + "step": 36731 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 36732 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 36733 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 36734 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 36735 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 36736 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 36737 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 36738 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 36739 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 36740 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 36741 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 36742 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.144, + "step": 36743 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 36744 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 36745 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 36746 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 36747 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 36748 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 36749 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 36750 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 36751 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 36752 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 36753 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 36754 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 36755 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1549, + "step": 36756 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 36757 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 36758 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 36759 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 36760 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 36761 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 36762 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 36763 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 36764 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 36765 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 36766 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.0952, + "step": 36767 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 36768 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 36769 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 36770 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 36771 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1462, + "step": 36772 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 36773 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 36774 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 36775 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 36776 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 36777 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 36778 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 36779 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 36780 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 36781 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1451, + "step": 36782 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 36783 + }, + { + "epoch": 2.23, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 36784 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 36785 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 36786 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 36787 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 36788 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 36789 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 36790 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 36791 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.0759, + "step": 36792 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 36793 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 36794 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 36795 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 36796 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1451, + "step": 36797 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 36798 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 36799 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 36800 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 36801 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 36802 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 36803 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.0763, + "step": 36804 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 36805 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 36806 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 36807 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.0868, + "step": 36808 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1718, + "step": 36809 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 36810 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 36811 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 36812 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 36813 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1429, + "step": 36814 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 36815 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 36816 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 36817 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 36818 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 36819 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 36820 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 36821 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 36822 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 36823 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 36824 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 36825 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 36826 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1717, + "step": 36827 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.0857, + "step": 36828 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 36829 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 36830 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 36831 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 36832 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 36833 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 36834 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 36835 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 36836 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 36837 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.0822, + "step": 36838 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 36839 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.0935, + "step": 36840 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 36841 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 36842 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 36843 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 36844 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 36845 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 36846 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.0874, + "step": 36847 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 36848 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 36849 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 36850 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 36851 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 36852 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.0952, + "step": 36853 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 36854 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 36855 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 36856 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 36857 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 36858 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 36859 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 36860 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 36861 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 36862 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1617, + "step": 36863 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 36864 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 36865 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 36866 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1518, + "step": 36867 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 36868 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 36869 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 36870 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 36871 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 36872 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1597, + "step": 36873 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 36874 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 36875 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 36876 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 36877 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 36878 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 36879 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 36880 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 36881 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 36882 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 36883 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.0869, + "step": 36884 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 36885 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 36886 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 36887 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 36888 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 36889 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 36890 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 36891 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 36892 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 36893 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 36894 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 36895 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 36896 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 36897 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 36898 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 36899 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.0857, + "step": 36900 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 36901 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 36902 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 36903 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 36904 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 36905 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 36906 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 36907 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 36908 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.0661, + "step": 36909 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 36910 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 36911 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 36912 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 36913 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 36914 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1495, + "step": 36915 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 36916 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 36917 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 36918 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 36919 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 36920 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.154, + "step": 36921 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 36922 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 36923 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 36924 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 36925 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 36926 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 36927 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.0666, + "step": 36928 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1597, + "step": 36929 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 36930 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 36931 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 36932 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 36933 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 36934 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 36935 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 36936 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 36937 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 36938 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 36939 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 36940 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 36941 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 36942 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 36943 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 36944 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 36945 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.0803, + "step": 36946 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 36947 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 36948 + }, + { + "epoch": 2.24, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 36949 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 36950 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 36951 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 36952 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 36953 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.0817, + "step": 36954 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 36955 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 36956 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 36957 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 36958 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 36959 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1566, + "step": 36960 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 36961 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 36962 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 36963 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 36964 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 36965 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 36966 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 36967 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 36968 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.0813, + "step": 36969 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.0864, + "step": 36970 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 36971 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 36972 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.0929, + "step": 36973 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 36974 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 36975 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 36976 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 36977 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 36978 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 36979 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 36980 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 36981 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.0856, + "step": 36982 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 36983 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.0818, + "step": 36984 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 36985 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 36986 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 36987 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 36988 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 36989 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 36990 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 36991 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 36992 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 36993 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 36994 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 36995 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.0898, + "step": 36996 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 36997 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 36998 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 36999 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 37000 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 37001 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 37002 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 37003 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 37004 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 37005 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 37006 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 37007 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 37008 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 37009 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 37010 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 37011 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1657, + "step": 37012 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1601, + "step": 37013 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 37014 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 37015 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 37016 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 37017 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 37018 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 37019 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 37020 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 37021 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.0697, + "step": 37022 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 37023 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 37024 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 37025 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 37026 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 37027 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 37028 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 37029 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 37030 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 37031 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 37032 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 37033 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 37034 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 37035 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.0901, + "step": 37036 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 37037 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 37038 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 37039 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1567, + "step": 37040 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 37041 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.0901, + "step": 37042 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 37043 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 37044 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 37045 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 37046 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 37047 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.0916, + "step": 37048 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 37049 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.0803, + "step": 37050 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 37051 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 37052 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 37053 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 37054 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 37055 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 37056 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 37057 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 37058 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 37059 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 37060 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1525, + "step": 37061 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 37062 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 37063 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 37064 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 37065 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 37066 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 37067 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 37068 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 37069 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 37070 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 37071 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 37072 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 37073 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 37074 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 37075 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 37076 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 37077 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 37078 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 37079 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 37080 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 37081 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 37082 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 37083 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 37084 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1635, + "step": 37085 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 37086 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 37087 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1634, + "step": 37088 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 37089 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 37090 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 37091 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 37092 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 37093 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 37094 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 37095 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 37096 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.0762, + "step": 37097 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 37098 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 37099 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 37100 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 37101 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.0917, + "step": 37102 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 37103 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 37104 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 37105 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 37106 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 37107 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 37108 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.0879, + "step": 37109 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 37110 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 37111 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 37112 + }, + { + "epoch": 2.25, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 37113 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 37114 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.0882, + "step": 37115 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 37116 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 37117 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 37118 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 37119 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 37120 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 37121 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 37122 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 37123 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 37124 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 37125 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 37126 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 37127 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.0876, + "step": 37128 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 37129 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 37130 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 37131 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 37132 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 37133 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 37134 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.0857, + "step": 37135 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 37136 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 37137 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.0765, + "step": 37138 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 37139 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 37140 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 37141 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 37142 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 37143 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 37144 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 37145 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.089, + "step": 37146 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 37147 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.085, + "step": 37148 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.0791, + "step": 37149 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 37150 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 37151 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 37152 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 37153 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 37154 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 37155 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 37156 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 37157 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 37158 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 37159 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 37160 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 37161 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 37162 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 37163 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1524, + "step": 37164 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 37165 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 37166 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 37167 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 37168 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 37169 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 37170 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 37171 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1603, + "step": 37172 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 37173 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1607, + "step": 37174 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 37175 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 37176 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 37177 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 37178 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 37179 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 37180 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 37181 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 37182 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 37183 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 37184 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1566, + "step": 37185 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 37186 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 37187 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 37188 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.0788, + "step": 37189 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 37190 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 37191 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 37192 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 37193 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.0932, + "step": 37194 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 37195 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1623, + "step": 37196 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 37197 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 37198 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 37199 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1635, + "step": 37200 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 37201 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 37202 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 37203 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 37204 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.0863, + "step": 37205 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 37206 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 37207 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 37208 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 37209 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 37210 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 37211 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 37212 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 37213 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 37214 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.0891, + "step": 37215 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 37216 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 37217 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 37218 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 37219 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 37220 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 37221 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 37222 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.0827, + "step": 37223 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 37224 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 37225 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 37226 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 37227 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.08, + "step": 37228 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 37229 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 37230 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 37231 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 37232 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 37233 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 37234 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 37235 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 37236 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 37237 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 37238 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 37239 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.0887, + "step": 37240 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 37241 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 37242 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 37243 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1528, + "step": 37244 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 37245 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 37246 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 37247 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 37248 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.0864, + "step": 37249 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1704, + "step": 37250 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 37251 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 37252 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 37253 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 37254 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 37255 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 37256 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 37257 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 37258 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 37259 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 37260 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 37261 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.147, + "step": 37262 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 37263 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 37264 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1767, + "step": 37265 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 37266 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 37267 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 37268 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1561, + "step": 37269 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 37270 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.0784, + "step": 37271 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 37272 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 37273 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 37274 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 37275 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 37276 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 37277 + }, + { + "epoch": 2.26, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 37278 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 37279 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 37280 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 37281 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 37282 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 37283 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 37284 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 37285 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 37286 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 37287 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 37288 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 37289 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 37290 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 37291 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 37292 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 37293 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.0783, + "step": 37294 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.0868, + "step": 37295 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 37296 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 37297 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.0872, + "step": 37298 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.0915, + "step": 37299 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 37300 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.085, + "step": 37301 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 37302 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.0851, + "step": 37303 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 37304 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 37305 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1453, + "step": 37306 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 37307 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 37308 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 37309 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 37310 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 37311 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.0788, + "step": 37312 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 37313 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 37314 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.0785, + "step": 37315 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 37316 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 37317 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 37318 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 37319 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 37320 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 37321 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 37322 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 37323 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 37324 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 37325 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 37326 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 37327 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 37328 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 37329 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 37330 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 37331 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 37332 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 37333 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.0899, + "step": 37334 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.083, + "step": 37335 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 37336 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 37337 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 37338 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 37339 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 37340 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 37341 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 37342 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 37343 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 37344 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 37345 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 37346 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1546, + "step": 37347 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 37348 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 37349 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 37350 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 37351 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 37352 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 37353 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 37354 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 37355 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.0886, + "step": 37356 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 37357 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 37358 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 37359 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 37360 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 37361 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 37362 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 37363 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 37364 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 37365 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 37366 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 37367 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 37368 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 37369 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 37370 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 37371 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 37372 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 37373 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 37374 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1488, + "step": 37375 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 37376 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 37377 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.0799, + "step": 37378 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 37379 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 37380 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 37381 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1689, + "step": 37382 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 37383 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 37384 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 37385 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 37386 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 37387 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 37388 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 37389 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 37390 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 37391 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 37392 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 37393 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 37394 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 37395 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 37396 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 37397 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 37398 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 37399 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 37400 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 37401 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 37402 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 37403 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 37404 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 37405 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1497, + "step": 37406 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 37407 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 37408 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 37409 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 37410 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 37411 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 37412 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 37413 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 37414 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 37415 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 37416 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 37417 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.0941, + "step": 37418 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 37419 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 37420 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 37421 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 37422 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 37423 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 37424 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 37425 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 37426 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1519, + "step": 37427 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 37428 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 37429 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 37430 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1681, + "step": 37431 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 37432 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1551, + "step": 37433 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 37434 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 37435 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 37436 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 37437 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 37438 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 37439 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 37440 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 37441 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 37442 + }, + { + "epoch": 2.27, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 37443 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 37444 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 37445 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 37446 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 37447 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 37448 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 37449 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 37450 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 37451 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 37452 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.0929, + "step": 37453 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 37454 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 37455 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 37456 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 37457 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 37458 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 37459 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 37460 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 37461 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 37462 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 37463 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 37464 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 37465 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 37466 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 37467 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 37468 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 37469 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 37470 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 37471 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 37472 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 37473 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.0852, + "step": 37474 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 37475 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 37476 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 37477 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 37478 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 37479 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 37480 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 37481 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 37482 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 37483 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 37484 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 37485 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.0888, + "step": 37486 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 37487 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.0886, + "step": 37488 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 37489 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 37490 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.0878, + "step": 37491 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 37492 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 37493 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 37494 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1598, + "step": 37495 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.0862, + "step": 37496 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 37497 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 37498 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 37499 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 37500 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 37501 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 37502 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 37503 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 37504 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 37505 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 37506 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 37507 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 37508 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 37509 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 37510 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 37511 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 37512 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 37513 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 37514 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 37515 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 37516 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 37517 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 37518 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 37519 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.0936, + "step": 37520 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.146, + "step": 37521 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1532, + "step": 37522 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 37523 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1564, + "step": 37524 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 37525 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 37526 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 37527 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 37528 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 37529 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 37530 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 37531 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 37532 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 37533 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.0943, + "step": 37534 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1468, + "step": 37535 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.162, + "step": 37536 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 37537 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 37538 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 37539 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 37540 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 37541 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1868, + "step": 37542 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 37543 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 37544 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 37545 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 37546 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 37547 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 37548 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 37549 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 37550 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 37551 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 37552 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 37553 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 37554 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 37555 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 37556 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 37557 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 37558 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 37559 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.0827, + "step": 37560 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 37561 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 37562 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 37563 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 37564 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 37565 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 37566 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 37567 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 37568 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 37569 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 37570 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 37571 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 37572 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 37573 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.0682, + "step": 37574 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 37575 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 37576 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 37577 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 37578 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 37579 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 37580 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 37581 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 37582 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.089, + "step": 37583 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 37584 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 37585 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 37586 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.0901, + "step": 37587 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 37588 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.0757, + "step": 37589 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 37590 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 37591 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 37592 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 37593 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.155, + "step": 37594 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 37595 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1727, + "step": 37596 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 37597 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.0917, + "step": 37598 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 37599 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 37600 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 37601 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 37602 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 37603 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 37604 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 37605 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.0828, + "step": 37606 + }, + { + "epoch": 2.28, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 37607 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 37608 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 37609 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 37610 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 37611 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 37612 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 37613 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1552, + "step": 37614 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 37615 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1586, + "step": 37616 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 37617 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 37618 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1521, + "step": 37619 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 37620 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 37621 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 37622 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 37623 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 37624 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 37625 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 37626 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 37627 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 37628 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 37629 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 37630 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 37631 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 37632 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 37633 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 37634 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 37635 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 37636 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1535, + "step": 37637 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 37638 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 37639 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 37640 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 37641 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1546, + "step": 37642 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 37643 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 37644 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 37645 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 37646 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 37647 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 37648 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 37649 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 37650 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 37651 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 37652 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.088, + "step": 37653 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 37654 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.069, + "step": 37655 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 37656 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 37657 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.0792, + "step": 37658 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 37659 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 37660 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 37661 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 37662 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 37663 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 37664 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 37665 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 37666 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 37667 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.0916, + "step": 37668 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 37669 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 37670 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 37671 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 37672 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 37673 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 37674 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 37675 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 37676 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 37677 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 37678 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.085, + "step": 37679 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 37680 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 37681 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 37682 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 37683 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 37684 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 37685 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 37686 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 37687 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 37688 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.0887, + "step": 37689 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 37690 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 37691 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 37692 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 37693 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 37694 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 37695 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 37696 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 37697 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 37698 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 37699 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 37700 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 37701 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.0935, + "step": 37702 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.0853, + "step": 37703 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 37704 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 37705 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 37706 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 37707 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 37708 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 37709 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 37710 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.0898, + "step": 37711 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 37712 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 37713 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 37714 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 37715 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 37716 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 37717 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 37718 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 37719 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 37720 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 37721 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 37722 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 37723 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 37724 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 37725 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 37726 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 37727 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 37728 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 37729 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 37730 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 37731 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 37732 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 37733 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 37734 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 37735 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 37736 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 37737 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 37738 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 37739 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 37740 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 37741 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 37742 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 37743 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.0847, + "step": 37744 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 37745 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.0782, + "step": 37746 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 37747 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 37748 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 37749 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 37750 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 37751 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 37752 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 37753 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 37754 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 37755 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 37756 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 37757 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 37758 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 37759 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 37760 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 37761 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 37762 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 37763 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 37764 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 37765 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 37766 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 37767 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 37768 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 37769 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 37770 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 37771 + }, + { + "epoch": 2.29, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 37772 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 37773 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 37774 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 37775 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 37776 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 37777 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 37778 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 37779 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 37780 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 37781 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 37782 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 37783 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 37784 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1453, + "step": 37785 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.0799, + "step": 37786 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 37787 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 37788 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 37789 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 37790 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.0869, + "step": 37791 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 37792 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 37793 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 37794 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 37795 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 37796 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 37797 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 37798 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 37799 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 37800 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 37801 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 37802 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 37803 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 37804 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 37805 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 37806 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 37807 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 37808 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 37809 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 37810 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 37811 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 37812 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 37813 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 37814 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1585, + "step": 37815 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.0859, + "step": 37816 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 37817 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 37818 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1524, + "step": 37819 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 37820 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 37821 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 37822 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 37823 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 37824 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 37825 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.0894, + "step": 37826 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 37827 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 37828 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 37829 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 37830 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 37831 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 37832 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 37833 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 37834 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 37835 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 37836 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 37837 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 37838 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 37839 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 37840 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 37841 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 37842 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 37843 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 37844 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 37845 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 37846 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 37847 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 37848 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 37849 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 37850 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 37851 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 37852 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 37853 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 37854 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 37855 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 37856 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 37857 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.0827, + "step": 37858 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 37859 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.145, + "step": 37860 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 37861 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 37862 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 37863 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 37864 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 37865 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 37866 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 37867 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 37868 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 37869 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 37870 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.091, + "step": 37871 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 37872 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 37873 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 37874 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 37875 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 37876 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 37877 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 37878 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 37879 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 37880 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 37881 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 37882 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 37883 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 37884 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 37885 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 37886 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 37887 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 37888 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.0903, + "step": 37889 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.0886, + "step": 37890 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 37891 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 37892 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 37893 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.0845, + "step": 37894 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 37895 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 37896 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 37897 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 37898 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 37899 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 37900 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 37901 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 37902 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 37903 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 37904 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.154, + "step": 37905 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 37906 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 37907 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 37908 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.0869, + "step": 37909 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 37910 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 37911 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 37912 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 37913 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 37914 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 37915 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 37916 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 37917 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.0866, + "step": 37918 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 37919 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 37920 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 37921 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 37922 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1691, + "step": 37923 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 37924 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 37925 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 37926 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 37927 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 37928 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 37929 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 37930 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 37931 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.0935, + "step": 37932 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 37933 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 37934 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 37935 + }, + { + "epoch": 2.3, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 37936 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 37937 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 37938 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.0703, + "step": 37939 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 37940 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 37941 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 37942 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 37943 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 37944 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 37945 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 37946 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 37947 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 37948 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 37949 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 37950 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 37951 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 37952 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 37953 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 37954 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 37955 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 37956 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 37957 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 37958 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 37959 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.0831, + "step": 37960 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 37961 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 37962 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 37963 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 37964 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 37965 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 37966 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 37967 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 37968 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.145, + "step": 37969 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 37970 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 37971 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 37972 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 37973 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 37974 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 37975 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 37976 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 37977 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 37978 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 37979 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 37980 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 37981 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 37982 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 37983 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 37984 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 37985 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 37986 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 37987 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 37988 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 37989 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 37990 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 37991 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 37992 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 37993 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 37994 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 37995 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 37996 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 37997 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 37998 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 37999 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 38000 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 38001 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 38002 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 38003 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 38004 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 38005 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 38006 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 38007 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 38008 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 38009 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 38010 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 38011 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.0825, + "step": 38012 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 38013 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 38014 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 38015 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 38016 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 38017 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 38018 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 38019 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 38020 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 38021 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 38022 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 38023 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 38024 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.0941, + "step": 38025 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1515, + "step": 38026 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 38027 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 38028 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 38029 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 38030 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 38031 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 38032 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 38033 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 38034 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 38035 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 38036 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 38037 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 38038 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 38039 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 38040 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 38041 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 38042 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 38043 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 38044 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 38045 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 38046 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 38047 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 38048 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 38049 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 38050 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 38051 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 38052 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 38053 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 38054 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 38055 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 38056 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.088, + "step": 38057 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 38058 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.16, + "step": 38059 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 38060 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 38061 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.0898, + "step": 38062 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 38063 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 38064 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 38065 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 38066 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 38067 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 38068 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 38069 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 38070 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 38071 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 38072 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 38073 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 38074 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 38075 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 38076 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 38077 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 38078 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 38079 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 38080 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 38081 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.0941, + "step": 38082 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 38083 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 38084 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 38085 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 38086 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 38087 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 38088 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 38089 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 38090 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 38091 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 38092 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 38093 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 38094 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.0869, + "step": 38095 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 38096 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 38097 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 38098 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 38099 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 38100 + }, + { + "epoch": 2.31, + "learning_rate": 2e-05, + "loss": 0.1638, + "step": 38101 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 38102 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 38103 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 38104 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 38105 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 38106 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1778, + "step": 38107 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 38108 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 38109 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 38110 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 38111 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 38112 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 38113 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 38114 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 38115 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 38116 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 38117 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 38118 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 38119 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.0832, + "step": 38120 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 38121 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 38122 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 38123 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1675, + "step": 38124 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 38125 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 38126 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 38127 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 38128 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 38129 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 38130 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 38131 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 38132 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 38133 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 38134 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 38135 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 38136 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 38137 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 38138 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 38139 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.0849, + "step": 38140 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 38141 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 38142 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 38143 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 38144 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1679, + "step": 38145 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 38146 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 38147 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 38148 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 38149 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 38150 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 38151 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 38152 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 38153 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 38154 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1703, + "step": 38155 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 38156 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 38157 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 38158 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 38159 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 38160 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 38161 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 38162 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 38163 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1613, + "step": 38164 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 38165 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 38166 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 38167 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 38168 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 38169 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 38170 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 38171 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 38172 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 38173 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 38174 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 38175 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 38176 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 38177 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 38178 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 38179 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 38180 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 38181 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1796, + "step": 38182 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.0897, + "step": 38183 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 38184 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 38185 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 38186 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 38187 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.0943, + "step": 38188 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 38189 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 38190 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 38191 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 38192 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 38193 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 38194 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.0864, + "step": 38195 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 38196 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 38197 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 38198 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 38199 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 38200 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.0747, + "step": 38201 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 38202 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 38203 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 38204 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 38205 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 38206 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 38207 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 38208 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 38209 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 38210 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 38211 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 38212 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 38213 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 38214 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 38215 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 38216 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.153, + "step": 38217 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1503, + "step": 38218 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 38219 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 38220 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 38221 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 38222 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 38223 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 38224 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 38225 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 38226 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 38227 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 38228 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 38229 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 38230 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.0952, + "step": 38231 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 38232 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 38233 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 38234 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 38235 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 38236 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 38237 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.0722, + "step": 38238 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 38239 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 38240 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 38241 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 38242 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 38243 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1629, + "step": 38244 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.0786, + "step": 38245 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 38246 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 38247 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1465, + "step": 38248 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 38249 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 38250 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 38251 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 38252 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 38253 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 38254 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 38255 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.0899, + "step": 38256 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 38257 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 38258 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 38259 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 38260 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 38261 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 38262 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 38263 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 38264 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 38265 + }, + { + "epoch": 2.32, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 38266 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 38267 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 38268 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 38269 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 38270 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 38271 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 38272 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 38273 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 38274 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.0935, + "step": 38275 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 38276 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 38277 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 38278 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 38279 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 38280 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1627, + "step": 38281 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 38282 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.0866, + "step": 38283 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 38284 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 38285 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 38286 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 38287 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 38288 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 38289 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 38290 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 38291 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 38292 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 38293 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 38294 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 38295 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 38296 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 38297 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 38298 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 38299 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 38300 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 38301 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 38302 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 38303 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 38304 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 38305 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 38306 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 38307 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 38308 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 38309 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 38310 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 38311 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 38312 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 38313 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 38314 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 38315 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 38316 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 38317 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 38318 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 38319 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.09, + "step": 38320 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 38321 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 38322 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 38323 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 38324 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 38325 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 38326 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 38327 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 38328 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 38329 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 38330 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 38331 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 38332 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 38333 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 38334 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 38335 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 38336 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 38337 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 38338 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 38339 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 38340 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 38341 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 38342 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 38343 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 38344 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 38345 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 38346 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 38347 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 38348 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 38349 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 38350 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 38351 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.0894, + "step": 38352 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 38353 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 38354 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 38355 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 38356 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 38357 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 38358 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 38359 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 38360 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 38361 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 38362 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.0806, + "step": 38363 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 38364 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.0795, + "step": 38365 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 38366 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 38367 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 38368 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1676, + "step": 38369 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 38370 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 38371 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 38372 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 38373 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 38374 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 38375 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 38376 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 38377 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 38378 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 38379 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 38380 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 38381 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.087, + "step": 38382 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 38383 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.0914, + "step": 38384 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 38385 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.0891, + "step": 38386 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 38387 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 38388 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 38389 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 38390 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 38391 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 38392 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 38393 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 38394 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 38395 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.0739, + "step": 38396 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 38397 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 38398 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 38399 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 38400 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 38401 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 38402 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 38403 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 38404 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 38405 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 38406 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 38407 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 38408 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 38409 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 38410 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 38411 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.0804, + "step": 38412 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 38413 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 38414 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 38415 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 38416 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 38417 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 38418 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 38419 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1706, + "step": 38420 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.082, + "step": 38421 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 38422 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 38423 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 38424 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 38425 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 38426 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 38427 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 38428 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 38429 + }, + { + "epoch": 2.33, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 38430 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 38431 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 38432 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 38433 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 38434 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 38435 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.0775, + "step": 38436 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.0739, + "step": 38437 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 38438 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.0824, + "step": 38439 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 38440 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1495, + "step": 38441 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 38442 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 38443 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 38444 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 38445 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 38446 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.0929, + "step": 38447 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 38448 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 38449 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 38450 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 38451 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 38452 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1576, + "step": 38453 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 38454 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 38455 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 38456 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 38457 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 38458 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 38459 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 38460 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 38461 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 38462 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 38463 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 38464 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 38465 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 38466 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 38467 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 38468 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 38469 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 38470 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 38471 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.0839, + "step": 38472 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 38473 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 38474 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.0889, + "step": 38475 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 38476 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 38477 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 38478 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 38479 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 38480 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 38481 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 38482 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.0823, + "step": 38483 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 38484 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 38485 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 38486 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 38487 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 38488 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 38489 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 38490 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.163, + "step": 38491 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 38492 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 38493 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 38494 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 38495 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 38496 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 38497 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 38498 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 38499 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 38500 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 38501 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 38502 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 38503 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 38504 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 38505 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 38506 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 38507 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 38508 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 38509 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 38510 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.0769, + "step": 38511 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 38512 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 38513 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 38514 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 38515 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 38516 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 38517 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 38518 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 38519 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 38520 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.0747, + "step": 38521 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 38522 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 38523 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 38524 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 38525 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1442, + "step": 38526 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 38527 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 38528 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 38529 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.0771, + "step": 38530 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 38531 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 38532 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 38533 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 38534 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 38535 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 38536 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 38537 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 38538 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 38539 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 38540 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 38541 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 38542 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 38543 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 38544 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 38545 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 38546 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 38547 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 38548 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 38549 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 38550 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 38551 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 38552 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 38553 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 38554 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 38555 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 38556 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 38557 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 38558 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 38559 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 38560 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 38561 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 38562 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 38563 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1476, + "step": 38564 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 38565 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 38566 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 38567 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 38568 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 38569 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 38570 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 38571 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 38572 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 38573 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 38574 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 38575 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 38576 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 38577 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.0916, + "step": 38578 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 38579 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 38580 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 38581 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 38582 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 38583 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 38584 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 38585 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 38586 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 38587 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 38588 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 38589 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 38590 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 38591 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.0858, + "step": 38592 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 38593 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 38594 + }, + { + "epoch": 2.34, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 38595 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.084, + "step": 38596 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 38597 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 38598 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.0888, + "step": 38599 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 38600 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 38601 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 38602 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 38603 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 38604 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 38605 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 38606 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1498, + "step": 38607 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 38608 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 38609 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 38610 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 38611 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 38612 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 38613 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 38614 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 38615 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 38616 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 38617 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 38618 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 38619 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.146, + "step": 38620 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 38621 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 38622 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 38623 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 38624 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 38625 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 38626 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 38627 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 38628 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 38629 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 38630 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 38631 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 38632 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 38633 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 38634 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 38635 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.0864, + "step": 38636 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1575, + "step": 38637 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 38638 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 38639 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 38640 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 38641 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 38642 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 38643 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 38644 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.163, + "step": 38645 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 38646 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.0731, + "step": 38647 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 38648 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1585, + "step": 38649 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 38650 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 38651 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 38652 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 38653 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 38654 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 38655 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.0887, + "step": 38656 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1517, + "step": 38657 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 38658 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.152, + "step": 38659 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 38660 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 38661 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 38662 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 38663 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 38664 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1485, + "step": 38665 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 38666 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.0866, + "step": 38667 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 38668 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.0909, + "step": 38669 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 38670 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1453, + "step": 38671 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 38672 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 38673 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 38674 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 38675 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 38676 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 38677 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 38678 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 38679 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 38680 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 38681 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 38682 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 38683 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 38684 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 38685 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 38686 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 38687 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 38688 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 38689 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 38690 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 38691 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 38692 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.0825, + "step": 38693 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 38694 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 38695 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 38696 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 38697 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 38698 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 38699 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.0816, + "step": 38700 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 38701 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 38702 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.0902, + "step": 38703 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 38704 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 38705 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 38706 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 38707 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 38708 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 38709 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 38710 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 38711 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 38712 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 38713 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.087, + "step": 38714 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 38715 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 38716 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 38717 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 38718 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 38719 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 38720 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 38721 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 38722 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 38723 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 38724 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 38725 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 38726 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 38727 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 38728 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 38729 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 38730 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 38731 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 38732 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 38733 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 38734 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.0857, + "step": 38735 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 38736 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 38737 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 38738 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 38739 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 38740 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 38741 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 38742 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 38743 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 38744 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 38745 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 38746 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 38747 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 38748 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 38749 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.0918, + "step": 38750 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 38751 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 38752 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 38753 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 38754 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 38755 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 38756 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 38757 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1582, + "step": 38758 + }, + { + "epoch": 2.35, + "learning_rate": 2e-05, + "loss": 0.1585, + "step": 38759 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 38760 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 38761 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 38762 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 38763 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 38764 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 38765 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 38766 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 38767 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 38768 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 38769 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 38770 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 38771 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 38772 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 38773 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 38774 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 38775 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 38776 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 38777 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 38778 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 38779 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 38780 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 38781 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 38782 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 38783 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 38784 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 38785 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 38786 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 38787 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 38788 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 38789 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 38790 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 38791 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 38792 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 38793 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 38794 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 38795 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1637, + "step": 38796 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 38797 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 38798 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 38799 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 38800 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 38801 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 38802 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 38803 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 38804 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 38805 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 38806 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 38807 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1462, + "step": 38808 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.0788, + "step": 38809 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 38810 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 38811 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.0829, + "step": 38812 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 38813 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 38814 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 38815 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 38816 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 38817 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.0837, + "step": 38818 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 38819 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.0693, + "step": 38820 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 38821 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.0841, + "step": 38822 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 38823 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 38824 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 38825 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 38826 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 38827 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 38828 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 38829 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 38830 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 38831 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 38832 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 38833 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 38834 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 38835 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 38836 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 38837 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 38838 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.0876, + "step": 38839 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 38840 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.089, + "step": 38841 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 38842 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.0736, + "step": 38843 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.08, + "step": 38844 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 38845 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1469, + "step": 38846 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 38847 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 38848 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.0967, + "step": 38849 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 38850 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 38851 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 38852 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 38853 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 38854 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 38855 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 38856 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 38857 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.0825, + "step": 38858 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 38859 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1468, + "step": 38860 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 38861 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 38862 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 38863 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 38864 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 38865 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.0915, + "step": 38866 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 38867 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 38868 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 38869 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 38870 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 38871 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 38872 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 38873 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.0862, + "step": 38874 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 38875 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 38876 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 38877 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 38878 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 38879 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 38880 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 38881 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 38882 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 38883 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 38884 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 38885 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 38886 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 38887 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 38888 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 38889 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 38890 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 38891 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 38892 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 38893 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 38894 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 38895 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 38896 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 38897 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.0829, + "step": 38898 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1554, + "step": 38899 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 38900 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 38901 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 38902 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 38903 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 38904 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 38905 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 38906 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 38907 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 38908 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 38909 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 38910 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 38911 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 38912 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1599, + "step": 38913 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 38914 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1469, + "step": 38915 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 38916 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 38917 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 38918 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 38919 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.159, + "step": 38920 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 38921 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 38922 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 38923 + }, + { + "epoch": 2.36, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 38924 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 38925 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 38926 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 38927 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 38928 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 38929 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 38930 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 38931 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 38932 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 38933 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 38934 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 38935 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 38936 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 38937 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 38938 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 38939 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 38940 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 38941 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 38942 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 38943 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1443, + "step": 38944 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 38945 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 38946 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 38947 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 38948 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 38949 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 38950 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 38951 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 38952 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 38953 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 38954 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1447, + "step": 38955 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 38956 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 38957 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 38958 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 38959 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 38960 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 38961 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 38962 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 38963 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 38964 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 38965 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 38966 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.0882, + "step": 38967 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 38968 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 38969 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 38970 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.0873, + "step": 38971 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 38972 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 38973 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1652, + "step": 38974 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 38975 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 38976 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 38977 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 38978 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 38979 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 38980 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 38981 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 38982 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1638, + "step": 38983 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 38984 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 38985 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1688, + "step": 38986 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 38987 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1611, + "step": 38988 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 38989 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 38990 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.0791, + "step": 38991 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 38992 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 38993 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 38994 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 38995 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 38996 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 38997 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 38998 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 38999 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 39000 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 39001 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 39002 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 39003 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 39004 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 39005 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 39006 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 39007 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 39008 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 39009 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 39010 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 39011 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 39012 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 39013 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1772, + "step": 39014 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 39015 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 39016 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 39017 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 39018 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 39019 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 39020 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 39021 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 39022 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 39023 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 39024 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 39025 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 39026 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 39027 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 39028 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 39029 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 39030 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.082, + "step": 39031 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 39032 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 39033 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 39034 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 39035 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 39036 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 39037 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 39038 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1605, + "step": 39039 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 39040 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 39041 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 39042 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 39043 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 39044 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 39045 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 39046 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.0868, + "step": 39047 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 39048 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 39049 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 39050 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 39051 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 39052 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 39053 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 39054 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 39055 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 39056 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 39057 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 39058 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 39059 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 39060 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 39061 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 39062 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.082, + "step": 39063 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 39064 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 39065 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 39066 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 39067 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.147, + "step": 39068 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 39069 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1451, + "step": 39070 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 39071 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 39072 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 39073 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 39074 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 39075 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 39076 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.144, + "step": 39077 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 39078 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.0887, + "step": 39079 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 39080 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 39081 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 39082 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 39083 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 39084 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 39085 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 39086 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 39087 + }, + { + "epoch": 2.37, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 39088 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 39089 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 39090 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 39091 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 39092 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 39093 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1549, + "step": 39094 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 39095 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 39096 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 39097 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 39098 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 39099 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 39100 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 39101 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 39102 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 39103 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.186, + "step": 39104 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1564, + "step": 39105 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 39106 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 39107 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 39108 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 39109 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 39110 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 39111 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 39112 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 39113 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 39114 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 39115 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 39116 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 39117 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 39118 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 39119 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 39120 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 39121 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 39122 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 39123 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.0869, + "step": 39124 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 39125 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 39126 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 39127 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 39128 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.165, + "step": 39129 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 39130 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 39131 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 39132 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 39133 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 39134 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 39135 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 39136 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 39137 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 39138 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 39139 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 39140 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 39141 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 39142 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 39143 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 39144 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 39145 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 39146 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 39147 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 39148 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 39149 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 39150 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 39151 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 39152 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 39153 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.0868, + "step": 39154 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 39155 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.0895, + "step": 39156 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.0831, + "step": 39157 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.0814, + "step": 39158 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 39159 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 39160 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.0868, + "step": 39161 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 39162 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 39163 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 39164 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 39165 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.0838, + "step": 39166 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 39167 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 39168 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.0802, + "step": 39169 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 39170 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 39171 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 39172 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.071, + "step": 39173 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 39174 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 39175 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 39176 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 39177 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 39178 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 39179 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 39180 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 39181 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 39182 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 39183 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 39184 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 39185 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 39186 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 39187 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 39188 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 39189 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 39190 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 39191 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 39192 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 39193 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 39194 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 39195 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1596, + "step": 39196 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 39197 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 39198 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 39199 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 39200 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 39201 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 39202 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 39203 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 39204 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 39205 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 39206 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 39207 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 39208 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 39209 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 39210 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 39211 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 39212 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 39213 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 39214 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 39215 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 39216 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 39217 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 39218 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 39219 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 39220 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 39221 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 39222 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.0786, + "step": 39223 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 39224 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 39225 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 39226 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 39227 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 39228 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 39229 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 39230 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 39231 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 39232 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 39233 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 39234 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 39235 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.0909, + "step": 39236 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1468, + "step": 39237 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 39238 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 39239 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 39240 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.0665, + "step": 39241 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 39242 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 39243 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 39244 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 39245 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 39246 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 39247 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1786, + "step": 39248 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 39249 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 39250 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 39251 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 39252 + }, + { + "epoch": 2.38, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 39253 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 39254 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 39255 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 39256 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 39257 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 39258 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 39259 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 39260 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 39261 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 39262 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 39263 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.0914, + "step": 39264 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.0941, + "step": 39265 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.091, + "step": 39266 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 39267 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 39268 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 39269 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 39270 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 39271 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 39272 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 39273 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 39274 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 39275 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 39276 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 39277 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 39278 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 39279 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 39280 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 39281 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 39282 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 39283 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 39284 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 39285 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 39286 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 39287 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 39288 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 39289 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 39290 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 39291 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 39292 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 39293 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 39294 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 39295 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 39296 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 39297 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 39298 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 39299 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 39300 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 39301 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 39302 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 39303 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 39304 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.0819, + "step": 39305 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.09, + "step": 39306 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.0886, + "step": 39307 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 39308 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 39309 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 39310 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 39311 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.0804, + "step": 39312 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 39313 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 39314 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 39315 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.0891, + "step": 39316 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 39317 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 39318 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 39319 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 39320 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 39321 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 39322 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 39323 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 39324 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 39325 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 39326 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 39327 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 39328 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 39329 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 39330 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 39331 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 39332 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1548, + "step": 39333 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 39334 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 39335 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 39336 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.0801, + "step": 39337 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 39338 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 39339 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.0917, + "step": 39340 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 39341 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 39342 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 39343 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 39344 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1703, + "step": 39345 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 39346 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 39347 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.147, + "step": 39348 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 39349 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 39350 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 39351 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 39352 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 39353 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 39354 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 39355 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 39356 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 39357 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 39358 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 39359 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 39360 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.0811, + "step": 39361 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 39362 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.0911, + "step": 39363 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 39364 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 39365 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 39366 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 39367 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 39368 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 39369 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 39370 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 39371 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 39372 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 39373 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 39374 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 39375 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 39376 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 39377 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 39378 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 39379 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 39380 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 39381 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.0867, + "step": 39382 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 39383 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 39384 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 39385 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 39386 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 39387 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 39388 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 39389 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 39390 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.0915, + "step": 39391 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 39392 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 39393 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 39394 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 39395 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 39396 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 39397 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 39398 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 39399 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 39400 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 39401 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.0879, + "step": 39402 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 39403 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 39404 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 39405 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 39406 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 39407 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 39408 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 39409 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 39410 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 39411 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1442, + "step": 39412 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 39413 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 39414 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 39415 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 39416 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 39417 + }, + { + "epoch": 2.39, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 39418 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 39419 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 39420 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 39421 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 39422 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.0952, + "step": 39423 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 39424 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.0876, + "step": 39425 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 39426 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1667, + "step": 39427 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 39428 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.0809, + "step": 39429 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 39430 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 39431 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 39432 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 39433 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 39434 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 39435 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 39436 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 39437 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 39438 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 39439 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 39440 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.0932, + "step": 39441 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 39442 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 39443 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 39444 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 39445 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 39446 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 39447 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.0821, + "step": 39448 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 39449 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 39450 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 39451 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 39452 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 39453 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 39454 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 39455 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 39456 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 39457 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 39458 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 39459 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 39460 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 39461 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 39462 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 39463 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 39464 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 39465 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 39466 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 39467 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1515, + "step": 39468 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 39469 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 39470 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 39471 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 39472 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 39473 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 39474 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 39475 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 39476 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 39477 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 39478 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.0863, + "step": 39479 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.0739, + "step": 39480 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 39481 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.0873, + "step": 39482 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 39483 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 39484 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 39485 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 39486 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 39487 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 39488 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 39489 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 39490 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 39491 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 39492 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 39493 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 39494 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 39495 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 39496 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 39497 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 39498 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 39499 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 39500 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 39501 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 39502 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 39503 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 39504 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 39505 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 39506 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 39507 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 39508 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 39509 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.0846, + "step": 39510 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 39511 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 39512 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 39513 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 39514 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 39515 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 39516 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 39517 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 39518 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 39519 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 39520 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.0764, + "step": 39521 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1694, + "step": 39522 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.0858, + "step": 39523 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 39524 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 39525 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 39526 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 39527 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 39528 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 39529 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 39530 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 39531 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 39532 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1566, + "step": 39533 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 39534 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 39535 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 39536 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 39537 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 39538 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 39539 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 39540 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 39541 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 39542 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 39543 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 39544 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 39545 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.0887, + "step": 39546 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 39547 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1485, + "step": 39548 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 39549 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 39550 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 39551 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 39552 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 39553 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 39554 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 39555 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 39556 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 39557 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 39558 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 39559 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.0909, + "step": 39560 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 39561 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 39562 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 39563 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 39564 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 39565 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 39566 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.0848, + "step": 39567 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1708, + "step": 39568 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 39569 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 39570 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 39571 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 39572 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 39573 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 39574 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 39575 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 39576 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 39577 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 39578 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 39579 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 39580 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 39581 + }, + { + "epoch": 2.4, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 39582 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 39583 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 39584 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 39585 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 39586 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 39587 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 39588 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 39589 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 39590 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 39591 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 39592 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 39593 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 39594 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 39595 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 39596 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1429, + "step": 39597 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.0892, + "step": 39598 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 39599 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 39600 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 39601 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 39602 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 39603 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 39604 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 39605 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.081, + "step": 39606 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 39607 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 39608 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 39609 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 39610 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 39611 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 39612 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 39613 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 39614 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 39615 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 39616 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.0918, + "step": 39617 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 39618 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 39619 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 39620 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 39621 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 39622 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 39623 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 39624 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 39625 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 39626 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 39627 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.0819, + "step": 39628 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 39629 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 39630 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 39631 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 39632 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.087, + "step": 39633 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 39634 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 39635 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 39636 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 39637 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 39638 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 39639 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 39640 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 39641 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1521, + "step": 39642 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 39643 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.091, + "step": 39644 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 39645 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 39646 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 39647 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 39648 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 39649 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.154, + "step": 39650 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 39651 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 39652 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 39653 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 39654 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 39655 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 39656 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 39657 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 39658 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 39659 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 39660 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 39661 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1469, + "step": 39662 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 39663 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 39664 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1561, + "step": 39665 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 39666 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1676, + "step": 39667 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 39668 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 39669 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 39670 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 39671 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 39672 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 39673 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 39674 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 39675 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 39676 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 39677 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 39678 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 39679 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.0823, + "step": 39680 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 39681 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 39682 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 39683 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 39684 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 39685 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 39686 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 39687 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 39688 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 39689 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 39690 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 39691 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 39692 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 39693 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 39694 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1667, + "step": 39695 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 39696 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 39697 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 39698 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 39699 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 39700 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 39701 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 39702 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 39703 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 39704 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 39705 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 39706 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 39707 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 39708 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 39709 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 39710 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 39711 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 39712 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 39713 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1515, + "step": 39714 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 39715 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 39716 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 39717 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 39718 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 39719 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 39720 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 39721 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 39722 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1586, + "step": 39723 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 39724 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.0823, + "step": 39725 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 39726 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 39727 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 39728 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 39729 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 39730 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 39731 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 39732 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 39733 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 39734 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 39735 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 39736 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 39737 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 39738 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 39739 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 39740 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 39741 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 39742 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 39743 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.0784, + "step": 39744 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 39745 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 39746 + }, + { + "epoch": 2.41, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 39747 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 39748 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 39749 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 39750 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 39751 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1697, + "step": 39752 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.084, + "step": 39753 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 39754 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 39755 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 39756 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 39757 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 39758 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 39759 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 39760 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 39761 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.089, + "step": 39762 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 39763 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 39764 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 39765 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.0897, + "step": 39766 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 39767 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 39768 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 39769 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 39770 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1532, + "step": 39771 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 39772 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 39773 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 39774 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 39775 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.156, + "step": 39776 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 39777 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1632, + "step": 39778 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 39779 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 39780 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 39781 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 39782 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 39783 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 39784 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 39785 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 39786 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 39787 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 39788 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 39789 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 39790 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 39791 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 39792 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 39793 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.0914, + "step": 39794 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 39795 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 39796 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 39797 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.0827, + "step": 39798 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 39799 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 39800 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 39801 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 39802 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 39803 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 39804 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 39805 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 39806 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 39807 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 39808 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 39809 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 39810 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 39811 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 39812 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 39813 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 39814 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 39815 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 39816 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 39817 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 39818 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 39819 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.0823, + "step": 39820 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1567, + "step": 39821 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 39822 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 39823 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 39824 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 39825 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 39826 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 39827 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 39828 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 39829 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 39830 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 39831 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 39832 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.0788, + "step": 39833 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 39834 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 39835 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 39836 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 39837 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 39838 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 39839 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.0843, + "step": 39840 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 39841 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1462, + "step": 39842 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 39843 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 39844 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 39845 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 39846 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 39847 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 39848 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 39849 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 39850 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 39851 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 39852 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 39853 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 39854 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 39855 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1533, + "step": 39856 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 39857 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 39858 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 39859 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 39860 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 39861 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 39862 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 39863 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 39864 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 39865 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 39866 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.0869, + "step": 39867 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 39868 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 39869 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 39870 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 39871 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1511, + "step": 39872 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 39873 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 39874 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.0845, + "step": 39875 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 39876 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 39877 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 39878 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 39879 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 39880 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 39881 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 39882 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 39883 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 39884 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 39885 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 39886 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 39887 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 39888 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 39889 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 39890 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1784, + "step": 39891 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 39892 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 39893 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 39894 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 39895 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.0874, + "step": 39896 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 39897 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 39898 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 39899 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 39900 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 39901 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 39902 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 39903 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 39904 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 39905 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 39906 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 39907 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 39908 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 39909 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 39910 + }, + { + "epoch": 2.42, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 39911 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1628, + "step": 39912 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 39913 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 39914 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 39915 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 39916 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 39917 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 39918 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1761, + "step": 39919 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 39920 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 39921 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1578, + "step": 39922 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1628, + "step": 39923 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 39924 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 39925 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.0915, + "step": 39926 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 39927 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 39928 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 39929 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 39930 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 39931 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 39932 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 39933 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 39934 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 39935 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 39936 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 39937 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 39938 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 39939 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 39940 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 39941 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 39942 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 39943 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 39944 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 39945 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 39946 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 39947 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 39948 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 39949 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 39950 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 39951 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 39952 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 39953 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 39954 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 39955 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 39956 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 39957 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 39958 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 39959 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 39960 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 39961 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 39962 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 39963 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 39964 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 39965 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 39966 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1499, + "step": 39967 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 39968 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 39969 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 39970 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 39971 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 39972 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 39973 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 39974 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 39975 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.145, + "step": 39976 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 39977 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 39978 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 39979 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 39980 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 39981 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 39982 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 39983 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.0903, + "step": 39984 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 39985 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 39986 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 39987 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1606, + "step": 39988 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 39989 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 39990 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.0875, + "step": 39991 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.0834, + "step": 39992 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 39993 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 39994 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 39995 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 39996 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 39997 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 39998 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 39999 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 40000 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 40001 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 40002 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 40003 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 40004 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.0704, + "step": 40005 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 40006 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 40007 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 40008 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 40009 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 40010 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.0888, + "step": 40011 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.0941, + "step": 40012 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 40013 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.0799, + "step": 40014 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 40015 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 40016 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 40017 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 40018 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 40019 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 40020 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.0874, + "step": 40021 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 40022 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 40023 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 40024 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 40025 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 40026 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 40027 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 40028 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 40029 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1601, + "step": 40030 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 40031 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 40032 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 40033 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1749, + "step": 40034 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 40035 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 40036 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 40037 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.0776, + "step": 40038 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 40039 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 40040 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 40041 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 40042 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 40043 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 40044 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 40045 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 40046 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 40047 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 40048 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.0897, + "step": 40049 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 40050 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 40051 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 40052 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 40053 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 40054 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.162, + "step": 40055 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 40056 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 40057 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 40058 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 40059 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.0877, + "step": 40060 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 40061 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 40062 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.0902, + "step": 40063 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 40064 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 40065 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.0834, + "step": 40066 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 40067 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.0901, + "step": 40068 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 40069 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.0909, + "step": 40070 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.0903, + "step": 40071 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.0844, + "step": 40072 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.0906, + "step": 40073 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 40074 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 40075 + }, + { + "epoch": 2.43, + "learning_rate": 2e-05, + "loss": 0.0901, + "step": 40076 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 40077 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 40078 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.0868, + "step": 40079 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.0915, + "step": 40080 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 40081 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 40082 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1495, + "step": 40083 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 40084 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 40085 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 40086 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 40087 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 40088 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 40089 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 40090 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 40091 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 40092 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 40093 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 40094 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1709, + "step": 40095 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 40096 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 40097 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 40098 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 40099 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 40100 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 40101 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 40102 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 40103 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1495, + "step": 40104 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 40105 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 40106 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 40107 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 40108 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 40109 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 40110 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 40111 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 40112 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 40113 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.0853, + "step": 40114 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 40115 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 40116 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 40117 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1515, + "step": 40118 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 40119 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 40120 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 40121 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 40122 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 40123 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 40124 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 40125 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 40126 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 40127 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.0876, + "step": 40128 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 40129 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 40130 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 40131 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 40132 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.0843, + "step": 40133 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 40134 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.0911, + "step": 40135 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 40136 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 40137 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 40138 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 40139 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 40140 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1748, + "step": 40141 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 40142 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.0929, + "step": 40143 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 40144 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 40145 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 40146 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 40147 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 40148 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.0873, + "step": 40149 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 40150 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 40151 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 40152 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.0867, + "step": 40153 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 40154 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 40155 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 40156 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 40157 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 40158 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 40159 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 40160 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 40161 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 40162 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 40163 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 40164 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 40165 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 40166 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 40167 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 40168 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 40169 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 40170 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 40171 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 40172 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 40173 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.0882, + "step": 40174 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 40175 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 40176 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 40177 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 40178 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 40179 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 40180 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 40181 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 40182 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 40183 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.0857, + "step": 40184 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.0881, + "step": 40185 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.0936, + "step": 40186 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.0899, + "step": 40187 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 40188 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 40189 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 40190 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 40191 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 40192 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 40193 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 40194 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 40195 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 40196 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 40197 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 40198 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 40199 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 40200 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 40201 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 40202 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 40203 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 40204 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1548, + "step": 40205 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 40206 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 40207 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 40208 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 40209 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 40210 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 40211 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 40212 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 40213 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 40214 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 40215 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 40216 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 40217 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 40218 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 40219 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 40220 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 40221 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 40222 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 40223 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 40224 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 40225 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 40226 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1599, + "step": 40227 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 40228 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1535, + "step": 40229 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 40230 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 40231 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.0932, + "step": 40232 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 40233 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 40234 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1526, + "step": 40235 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 40236 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 40237 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 40238 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 40239 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 40240 + }, + { + "epoch": 2.44, + "learning_rate": 2e-05, + "loss": 0.0842, + "step": 40241 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1566, + "step": 40242 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 40243 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 40244 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 40245 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 40246 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 40247 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 40248 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 40249 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 40250 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 40251 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 40252 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1453, + "step": 40253 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 40254 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 40255 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 40256 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 40257 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 40258 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.0757, + "step": 40259 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 40260 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 40261 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 40262 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 40263 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 40264 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 40265 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 40266 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 40267 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 40268 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 40269 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 40270 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 40271 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 40272 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 40273 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 40274 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 40275 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 40276 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 40277 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 40278 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.147, + "step": 40279 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.154, + "step": 40280 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 40281 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 40282 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 40283 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 40284 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 40285 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 40286 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 40287 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.0792, + "step": 40288 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 40289 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 40290 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 40291 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 40292 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 40293 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 40294 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 40295 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 40296 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 40297 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 40298 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 40299 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 40300 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 40301 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 40302 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 40303 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 40304 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 40305 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 40306 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 40307 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 40308 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 40309 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 40310 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 40311 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 40312 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 40313 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 40314 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 40315 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 40316 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1689, + "step": 40317 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 40318 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 40319 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 40320 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.0899, + "step": 40321 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 40322 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.0881, + "step": 40323 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 40324 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 40325 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 40326 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 40327 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 40328 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 40329 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 40330 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 40331 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 40332 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.0863, + "step": 40333 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 40334 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 40335 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 40336 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 40337 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 40338 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 40339 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 40340 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 40341 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 40342 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 40343 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 40344 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 40345 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 40346 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 40347 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 40348 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 40349 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 40350 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 40351 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 40352 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 40353 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 40354 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 40355 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.0831, + "step": 40356 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.0853, + "step": 40357 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 40358 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 40359 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 40360 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.0918, + "step": 40361 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 40362 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 40363 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 40364 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 40365 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 40366 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 40367 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 40368 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 40369 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 40370 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 40371 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1592, + "step": 40372 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.0855, + "step": 40373 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 40374 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 40375 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 40376 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 40377 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 40378 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 40379 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 40380 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 40381 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 40382 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 40383 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 40384 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 40385 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 40386 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1533, + "step": 40387 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 40388 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 40389 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 40390 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 40391 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 40392 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 40393 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 40394 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 40395 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 40396 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 40397 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 40398 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 40399 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 40400 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 40401 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 40402 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 40403 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 40404 + }, + { + "epoch": 2.45, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 40405 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 40406 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 40407 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 40408 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 40409 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 40410 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 40411 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 40412 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 40413 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 40414 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 40415 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 40416 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 40417 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 40418 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 40419 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 40420 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 40421 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 40422 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 40423 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 40424 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.0914, + "step": 40425 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 40426 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 40427 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 40428 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 40429 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 40430 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 40431 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 40432 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 40433 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 40434 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 40435 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.0888, + "step": 40436 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 40437 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 40438 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 40439 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 40440 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 40441 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 40442 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.0875, + "step": 40443 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 40444 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 40445 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 40446 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 40447 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 40448 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 40449 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 40450 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 40451 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 40452 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 40453 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 40454 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 40455 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 40456 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 40457 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 40458 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 40459 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 40460 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 40461 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 40462 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 40463 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 40464 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 40465 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 40466 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 40467 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 40468 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 40469 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 40470 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 40471 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 40472 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 40473 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 40474 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.0866, + "step": 40475 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 40476 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 40477 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.0886, + "step": 40478 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 40479 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 40480 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 40481 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1453, + "step": 40482 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 40483 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 40484 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 40485 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 40486 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 40487 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 40488 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 40489 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 40490 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 40491 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.0863, + "step": 40492 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 40493 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 40494 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 40495 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 40496 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.0874, + "step": 40497 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 40498 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 40499 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 40500 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 40501 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.0572, + "step": 40502 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 40503 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 40504 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 40505 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 40506 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 40507 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1751, + "step": 40508 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 40509 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 40510 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 40511 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 40512 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 40513 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 40514 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 40515 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 40516 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.0835, + "step": 40517 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 40518 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 40519 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 40520 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 40521 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 40522 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 40523 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 40524 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 40525 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 40526 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 40527 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 40528 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 40529 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 40530 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 40531 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 40532 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 40533 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 40534 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 40535 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 40536 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 40537 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 40538 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.0815, + "step": 40539 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 40540 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 40541 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 40542 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 40543 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 40544 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 40545 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.0837, + "step": 40546 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 40547 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 40548 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 40549 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.0869, + "step": 40550 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1567, + "step": 40551 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.0902, + "step": 40552 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 40553 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 40554 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 40555 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 40556 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 40557 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 40558 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 40559 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 40560 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 40561 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 40562 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 40563 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 40564 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 40565 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 40566 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 40567 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 40568 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 40569 + }, + { + "epoch": 2.46, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 40570 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 40571 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 40572 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 40573 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1443, + "step": 40574 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 40575 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 40576 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 40577 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 40578 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 40579 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 40580 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 40581 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 40582 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 40583 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 40584 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 40585 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 40586 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 40587 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 40588 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 40589 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 40590 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 40591 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 40592 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 40593 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 40594 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 40595 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1554, + "step": 40596 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 40597 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 40598 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.0888, + "step": 40599 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 40600 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 40601 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 40602 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 40603 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 40604 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 40605 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 40606 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 40607 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.0784, + "step": 40608 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 40609 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 40610 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 40611 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 40612 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 40613 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 40614 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 40615 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 40616 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 40617 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 40618 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 40619 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.0909, + "step": 40620 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 40621 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 40622 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 40623 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 40624 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 40625 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 40626 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 40627 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 40628 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.159, + "step": 40629 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 40630 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 40631 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 40632 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 40633 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 40634 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1652, + "step": 40635 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.0864, + "step": 40636 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 40637 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 40638 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 40639 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 40640 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 40641 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 40642 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 40643 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 40644 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 40645 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 40646 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 40647 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 40648 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 40649 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 40650 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 40651 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 40652 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 40653 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 40654 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 40655 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 40656 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 40657 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 40658 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1672, + "step": 40659 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 40660 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 40661 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 40662 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 40663 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 40664 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 40665 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 40666 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 40667 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 40668 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 40669 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 40670 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 40671 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 40672 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 40673 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.0941, + "step": 40674 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 40675 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.0897, + "step": 40676 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 40677 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 40678 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1524, + "step": 40679 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 40680 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 40681 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 40682 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 40683 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.147, + "step": 40684 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 40685 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 40686 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 40687 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 40688 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 40689 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.0889, + "step": 40690 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 40691 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 40692 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 40693 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 40694 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 40695 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 40696 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 40697 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.0903, + "step": 40698 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 40699 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 40700 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 40701 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 40702 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 40703 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 40704 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 40705 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 40706 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 40707 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 40708 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 40709 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 40710 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 40711 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 40712 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 40713 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.0899, + "step": 40714 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 40715 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 40716 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 40717 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 40718 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 40719 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 40720 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.084, + "step": 40721 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 40722 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 40723 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.144, + "step": 40724 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 40725 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.0902, + "step": 40726 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 40727 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 40728 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 40729 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 40730 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1928, + "step": 40731 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 40732 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 40733 + }, + { + "epoch": 2.47, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 40734 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 40735 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 40736 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 40737 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 40738 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 40739 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 40740 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 40741 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.0815, + "step": 40742 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 40743 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 40744 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1478, + "step": 40745 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 40746 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 40747 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 40748 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 40749 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 40750 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 40751 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 40752 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 40753 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 40754 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 40755 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 40756 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 40757 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 40758 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 40759 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 40760 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 40761 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 40762 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1521, + "step": 40763 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1588, + "step": 40764 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 40765 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 40766 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 40767 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.0732, + "step": 40768 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 40769 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 40770 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 40771 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 40772 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 40773 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 40774 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 40775 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 40776 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 40777 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 40778 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 40779 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 40780 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 40781 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.088, + "step": 40782 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 40783 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 40784 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 40785 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 40786 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.0688, + "step": 40787 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 40788 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 40789 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1648, + "step": 40790 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 40791 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 40792 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.0846, + "step": 40793 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 40794 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 40795 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 40796 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 40797 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1518, + "step": 40798 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 40799 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 40800 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 40801 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 40802 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 40803 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.0797, + "step": 40804 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 40805 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 40806 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 40807 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 40808 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 40809 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 40810 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.091, + "step": 40811 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 40812 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.0899, + "step": 40813 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 40814 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 40815 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 40816 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 40817 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 40818 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 40819 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1502, + "step": 40820 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 40821 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1561, + "step": 40822 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 40823 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 40824 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 40825 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 40826 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 40827 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 40828 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 40829 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 40830 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 40831 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 40832 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 40833 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 40834 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 40835 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 40836 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 40837 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 40838 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 40839 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 40840 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.0914, + "step": 40841 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 40842 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 40843 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 40844 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.0848, + "step": 40845 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 40846 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.0883, + "step": 40847 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 40848 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 40849 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 40850 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 40851 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.144, + "step": 40852 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 40853 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 40854 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 40855 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 40856 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.0903, + "step": 40857 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 40858 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 40859 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 40860 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 40861 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 40862 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 40863 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 40864 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 40865 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 40866 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 40867 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 40868 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 40869 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 40870 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 40871 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 40872 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 40873 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.0842, + "step": 40874 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 40875 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 40876 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 40877 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 40878 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 40879 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 40880 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 40881 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.0917, + "step": 40882 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 40883 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 40884 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 40885 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 40886 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 40887 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 40888 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 40889 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 40890 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 40891 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 40892 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 40893 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1498, + "step": 40894 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 40895 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 40896 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 40897 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 40898 + }, + { + "epoch": 2.48, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 40899 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.0895, + "step": 40900 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 40901 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 40902 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 40903 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 40904 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 40905 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 40906 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 40907 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 40908 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 40909 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 40910 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 40911 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 40912 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 40913 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.159, + "step": 40914 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 40915 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 40916 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 40917 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 40918 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1456, + "step": 40919 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 40920 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 40921 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 40922 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 40923 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 40924 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 40925 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 40926 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 40927 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 40928 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 40929 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 40930 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 40931 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 40932 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 40933 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1626, + "step": 40934 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 40935 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 40936 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 40937 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 40938 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.145, + "step": 40939 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 40940 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 40941 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1664, + "step": 40942 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 40943 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1618, + "step": 40944 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.0842, + "step": 40945 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 40946 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 40947 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 40948 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 40949 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 40950 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 40951 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 40952 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 40953 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.0721, + "step": 40954 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 40955 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 40956 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.0828, + "step": 40957 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 40958 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 40959 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 40960 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 40961 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 40962 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 40963 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.0901, + "step": 40964 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 40965 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 40966 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 40967 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 40968 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 40969 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1585, + "step": 40970 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 40971 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 40972 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 40973 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1687, + "step": 40974 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 40975 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 40976 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 40977 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 40978 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 40979 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 40980 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 40981 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.0823, + "step": 40982 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 40983 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 40984 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.0863, + "step": 40985 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 40986 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 40987 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 40988 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 40989 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 40990 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 40991 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 40992 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 40993 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 40994 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 40995 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 40996 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 40997 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 40998 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 40999 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 41000 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 41001 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 41002 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.0755, + "step": 41003 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 41004 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 41005 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 41006 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 41007 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 41008 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 41009 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 41010 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 41011 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 41012 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.09, + "step": 41013 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 41014 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 41015 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 41016 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 41017 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 41018 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1579, + "step": 41019 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 41020 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 41021 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 41022 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 41023 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 41024 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 41025 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 41026 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 41027 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 41028 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 41029 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 41030 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 41031 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 41032 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.0877, + "step": 41033 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 41034 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 41035 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 41036 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 41037 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 41038 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.0916, + "step": 41039 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 41040 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 41041 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 41042 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 41043 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.0891, + "step": 41044 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 41045 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 41046 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 41047 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 41048 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 41049 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 41050 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 41051 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.0741, + "step": 41052 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 41053 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 41054 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 41055 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 41056 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 41057 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 41058 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.0911, + "step": 41059 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 41060 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 41061 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 41062 + }, + { + "epoch": 2.49, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 41063 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 41064 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 41065 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 41066 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 41067 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 41068 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 41069 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 41070 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 41071 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 41072 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 41073 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 41074 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 41075 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1502, + "step": 41076 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 41077 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1555, + "step": 41078 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 41079 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.0916, + "step": 41080 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 41081 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1497, + "step": 41082 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 41083 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 41084 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 41085 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 41086 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 41087 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 41088 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 41089 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 41090 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 41091 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 41092 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 41093 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.0911, + "step": 41094 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 41095 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 41096 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 41097 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 41098 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 41099 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 41100 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.0898, + "step": 41101 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 41102 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 41103 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 41104 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 41105 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 41106 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 41107 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 41108 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 41109 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 41110 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 41111 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 41112 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 41113 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 41114 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 41115 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 41116 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 41117 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.0863, + "step": 41118 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 41119 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 41120 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 41121 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 41122 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 41123 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 41124 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 41125 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 41126 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 41127 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 41128 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 41129 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 41130 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 41131 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 41132 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 41133 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 41134 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 41135 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.0936, + "step": 41136 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 41137 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 41138 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 41139 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1498, + "step": 41140 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 41141 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 41142 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 41143 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 41144 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 41145 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1673, + "step": 41146 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 41147 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 41148 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 41149 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 41150 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 41151 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 41152 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 41153 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 41154 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1429, + "step": 41155 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 41156 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 41157 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 41158 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 41159 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 41160 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 41161 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.0768, + "step": 41162 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 41163 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 41164 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 41165 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 41166 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 41167 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 41168 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 41169 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 41170 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 41171 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 41172 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1689, + "step": 41173 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 41174 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1592, + "step": 41175 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 41176 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 41177 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 41178 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 41179 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 41180 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 41181 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 41182 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 41183 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 41184 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 41185 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 41186 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 41187 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 41188 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 41189 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 41190 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 41191 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 41192 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 41193 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 41194 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 41195 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 41196 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 41197 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 41198 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 41199 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 41200 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 41201 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 41202 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 41203 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 41204 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 41205 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 41206 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.0929, + "step": 41207 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 41208 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 41209 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 41210 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 41211 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 41212 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 41213 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 41214 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 41215 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 41216 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 41217 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 41218 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 41219 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.0798, + "step": 41220 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 41221 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 41222 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 41223 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 41224 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 41225 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 41226 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 41227 + }, + { + "epoch": 2.5, + "learning_rate": 2e-05, + "loss": 0.0952, + "step": 41228 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 41229 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 41230 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 41231 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 41232 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 41233 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1505, + "step": 41234 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.0967, + "step": 41235 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 41236 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 41237 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.0936, + "step": 41238 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1663, + "step": 41239 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 41240 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 41241 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 41242 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 41243 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 41244 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 41245 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 41246 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 41247 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 41248 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 41249 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 41250 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 41251 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 41252 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 41253 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 41254 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 41255 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 41256 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 41257 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 41258 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 41259 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1628, + "step": 41260 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 41261 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 41262 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 41263 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 41264 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.076, + "step": 41265 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 41266 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1652, + "step": 41267 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 41268 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 41269 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 41270 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 41271 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 41272 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 41273 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 41274 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 41275 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 41276 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 41277 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 41278 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 41279 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1429, + "step": 41280 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 41281 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 41282 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 41283 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 41284 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 41285 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 41286 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 41287 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1691, + "step": 41288 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 41289 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 41290 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 41291 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 41292 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 41293 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 41294 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 41295 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.0873, + "step": 41296 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 41297 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 41298 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 41299 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 41300 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 41301 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 41302 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 41303 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 41304 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 41305 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.0853, + "step": 41306 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 41307 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 41308 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 41309 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 41310 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 41311 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 41312 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 41313 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 41314 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 41315 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 41316 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1726, + "step": 41317 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 41318 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 41319 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.0828, + "step": 41320 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 41321 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 41322 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 41323 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 41324 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 41325 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 41326 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 41327 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 41328 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1507, + "step": 41329 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 41330 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 41331 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 41332 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 41333 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 41334 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 41335 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.0769, + "step": 41336 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 41337 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 41338 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 41339 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 41340 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 41341 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1522, + "step": 41342 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.0686, + "step": 41343 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.0859, + "step": 41344 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 41345 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 41346 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 41347 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 41348 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 41349 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 41350 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 41351 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 41352 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 41353 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 41354 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 41355 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 41356 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 41357 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 41358 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 41359 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 41360 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.0875, + "step": 41361 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 41362 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 41363 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 41364 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1478, + "step": 41365 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 41366 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 41367 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 41368 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 41369 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 41370 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.0967, + "step": 41371 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 41372 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 41373 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 41374 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 41375 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 41376 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 41377 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.0967, + "step": 41378 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 41379 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 41380 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 41381 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 41382 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 41383 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 41384 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 41385 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 41386 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 41387 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 41388 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 41389 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 41390 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 41391 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 41392 + }, + { + "epoch": 2.51, + "learning_rate": 2e-05, + "loss": 0.0891, + "step": 41393 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 41394 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 41395 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 41396 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1616, + "step": 41397 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 41398 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 41399 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 41400 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 41401 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 41402 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 41403 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 41404 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.0841, + "step": 41405 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 41406 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 41407 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 41408 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 41409 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 41410 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 41411 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 41412 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 41413 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.0751, + "step": 41414 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 41415 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 41416 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1511, + "step": 41417 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 41418 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 41419 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 41420 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 41421 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 41422 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.0851, + "step": 41423 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 41424 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 41425 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 41426 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 41427 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 41428 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 41429 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 41430 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.0916, + "step": 41431 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 41432 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 41433 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.0809, + "step": 41434 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 41435 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 41436 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 41437 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.0854, + "step": 41438 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 41439 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 41440 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 41441 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1478, + "step": 41442 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 41443 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 41444 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 41445 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 41446 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 41447 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 41448 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 41449 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 41450 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 41451 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 41452 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 41453 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 41454 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 41455 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 41456 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 41457 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 41458 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 41459 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 41460 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.0855, + "step": 41461 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 41462 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 41463 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 41464 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 41465 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 41466 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 41467 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 41468 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 41469 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 41470 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 41471 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 41472 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 41473 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 41474 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 41475 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 41476 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 41477 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 41478 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 41479 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 41480 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 41481 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 41482 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 41483 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 41484 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 41485 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 41486 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 41487 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 41488 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 41489 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 41490 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 41491 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 41492 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 41493 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 41494 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 41495 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 41496 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 41497 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.0834, + "step": 41498 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 41499 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 41500 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 41501 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 41502 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 41503 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 41504 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 41505 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 41506 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.0863, + "step": 41507 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 41508 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.0837, + "step": 41509 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 41510 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 41511 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 41512 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 41513 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 41514 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.0758, + "step": 41515 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 41516 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 41517 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 41518 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 41519 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 41520 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.079, + "step": 41521 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 41522 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 41523 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.0936, + "step": 41524 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 41525 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.0793, + "step": 41526 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 41527 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.0797, + "step": 41528 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 41529 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 41530 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 41531 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.0916, + "step": 41532 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 41533 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 41534 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 41535 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 41536 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1551, + "step": 41537 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1656, + "step": 41538 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.0852, + "step": 41539 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 41540 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 41541 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 41542 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 41543 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 41544 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 41545 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 41546 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 41547 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 41548 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 41549 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 41550 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 41551 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.0967, + "step": 41552 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 41553 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 41554 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 41555 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.0865, + "step": 41556 + }, + { + "epoch": 2.52, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 41557 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 41558 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 41559 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 41560 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.146, + "step": 41561 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 41562 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 41563 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 41564 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 41565 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 41566 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 41567 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 41568 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.0887, + "step": 41569 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 41570 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 41571 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 41572 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 41573 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.081, + "step": 41574 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 41575 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 41576 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 41577 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.0776, + "step": 41578 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 41579 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 41580 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 41581 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 41582 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 41583 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 41584 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 41585 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 41586 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 41587 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 41588 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 41589 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 41590 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 41591 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 41592 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 41593 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 41594 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1515, + "step": 41595 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 41596 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 41597 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1721, + "step": 41598 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 41599 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 41600 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 41601 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 41602 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 41603 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 41604 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 41605 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 41606 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 41607 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 41608 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 41609 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 41610 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 41611 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 41612 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 41613 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 41614 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 41615 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 41616 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 41617 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 41618 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 41619 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 41620 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 41621 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1554, + "step": 41622 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 41623 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 41624 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 41625 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 41626 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 41627 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 41628 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1574, + "step": 41629 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 41630 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 41631 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 41632 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 41633 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 41634 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 41635 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 41636 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 41637 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 41638 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 41639 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 41640 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1451, + "step": 41641 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 41642 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 41643 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 41644 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 41645 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.0763, + "step": 41646 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 41647 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 41648 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 41649 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 41650 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 41651 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.0917, + "step": 41652 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 41653 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.0899, + "step": 41654 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 41655 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 41656 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.073, + "step": 41657 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 41658 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 41659 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 41660 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 41661 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 41662 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 41663 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 41664 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 41665 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 41666 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 41667 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 41668 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1558, + "step": 41669 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 41670 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 41671 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 41672 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 41673 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 41674 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 41675 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 41676 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 41677 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 41678 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 41679 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 41680 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 41681 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 41682 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 41683 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 41684 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 41685 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 41686 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 41687 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 41688 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 41689 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 41690 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1577, + "step": 41691 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.161, + "step": 41692 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 41693 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 41694 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 41695 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 41696 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 41697 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 41698 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 41699 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 41700 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 41701 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 41702 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 41703 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 41704 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 41705 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 41706 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 41707 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 41708 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 41709 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 41710 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 41711 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 41712 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 41713 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 41714 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 41715 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.0888, + "step": 41716 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 41717 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 41718 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.0878, + "step": 41719 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.0935, + "step": 41720 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 41721 + }, + { + "epoch": 2.53, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 41722 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 41723 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 41724 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 41725 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.0818, + "step": 41726 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 41727 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 41728 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 41729 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 41730 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 41731 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.0716, + "step": 41732 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 41733 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 41734 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 41735 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 41736 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 41737 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 41738 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 41739 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 41740 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.081, + "step": 41741 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 41742 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 41743 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 41744 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 41745 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 41746 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 41747 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 41748 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 41749 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.0702, + "step": 41750 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 41751 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 41752 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 41753 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.091, + "step": 41754 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 41755 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 41756 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 41757 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 41758 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 41759 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 41760 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 41761 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1465, + "step": 41762 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 41763 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 41764 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 41765 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 41766 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 41767 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 41768 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 41769 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 41770 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 41771 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1503, + "step": 41772 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.154, + "step": 41773 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 41774 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 41775 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.0891, + "step": 41776 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 41777 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 41778 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 41779 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 41780 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 41781 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 41782 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.0917, + "step": 41783 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 41784 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 41785 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 41786 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 41787 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 41788 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 41789 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 41790 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 41791 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 41792 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 41793 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.0906, + "step": 41794 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.0824, + "step": 41795 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 41796 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 41797 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 41798 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 41799 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 41800 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 41801 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 41802 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 41803 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 41804 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 41805 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 41806 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 41807 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 41808 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 41809 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 41810 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 41811 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 41812 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 41813 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 41814 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 41815 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 41816 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 41817 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.0814, + "step": 41818 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 41819 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 41820 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 41821 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 41822 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.0915, + "step": 41823 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.0853, + "step": 41824 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 41825 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 41826 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 41827 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 41828 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 41829 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 41830 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 41831 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 41832 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 41833 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 41834 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 41835 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1609, + "step": 41836 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 41837 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 41838 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 41839 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 41840 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 41841 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 41842 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 41843 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 41844 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 41845 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 41846 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 41847 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 41848 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 41849 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 41850 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 41851 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 41852 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 41853 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 41854 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 41855 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 41856 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 41857 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.0892, + "step": 41858 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 41859 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 41860 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 41861 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 41862 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.0718, + "step": 41863 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 41864 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 41865 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1499, + "step": 41866 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 41867 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.0638, + "step": 41868 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 41869 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 41870 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.0701, + "step": 41871 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 41872 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 41873 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 41874 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 41875 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 41876 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 41877 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 41878 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 41879 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 41880 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 41881 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 41882 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 41883 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 41884 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 41885 + }, + { + "epoch": 2.54, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 41886 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 41887 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 41888 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 41889 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 41890 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 41891 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1535, + "step": 41892 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 41893 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.0814, + "step": 41894 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 41895 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 41896 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 41897 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 41898 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 41899 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 41900 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.0858, + "step": 41901 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 41902 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 41903 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 41904 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 41905 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 41906 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 41907 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 41908 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 41909 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 41910 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 41911 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 41912 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 41913 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 41914 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 41915 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 41916 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 41917 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 41918 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 41919 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 41920 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.0906, + "step": 41921 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 41922 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 41923 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 41924 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 41925 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 41926 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 41927 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 41928 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 41929 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 41930 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 41931 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 41932 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 41933 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 41934 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 41935 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 41936 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 41937 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 41938 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 41939 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 41940 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 41941 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 41942 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 41943 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 41944 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 41945 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 41946 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 41947 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 41948 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 41949 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 41950 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 41951 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 41952 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 41953 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 41954 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 41955 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.0941, + "step": 41956 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.0895, + "step": 41957 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 41958 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 41959 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 41960 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 41961 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 41962 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 41963 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 41964 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 41965 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 41966 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 41967 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 41968 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.0869, + "step": 41969 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 41970 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 41971 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.0967, + "step": 41972 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.0888, + "step": 41973 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 41974 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.0879, + "step": 41975 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1527, + "step": 41976 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.0883, + "step": 41977 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 41978 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.0818, + "step": 41979 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 41980 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 41981 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 41982 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.0898, + "step": 41983 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 41984 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 41985 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 41986 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 41987 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 41988 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 41989 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 41990 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 41991 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 41992 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 41993 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 41994 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 41995 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 41996 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 41997 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 41998 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 41999 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 42000 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1639, + "step": 42001 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 42002 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 42003 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 42004 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 42005 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 42006 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 42007 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 42008 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1567, + "step": 42009 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 42010 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 42011 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 42012 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 42013 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 42014 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1582, + "step": 42015 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 42016 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 42017 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 42018 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 42019 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 42020 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 42021 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 42022 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 42023 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 42024 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 42025 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 42026 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 42027 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 42028 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 42029 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 42030 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 42031 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 42032 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.173, + "step": 42033 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 42034 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 42035 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.0826, + "step": 42036 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 42037 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 42038 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 42039 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 42040 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 42041 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 42042 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 42043 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 42044 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 42045 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 42046 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 42047 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 42048 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 42049 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 42050 + }, + { + "epoch": 2.55, + "learning_rate": 2e-05, + "loss": 0.1465, + "step": 42051 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 42052 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 42053 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 42054 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 42055 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 42056 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1585, + "step": 42057 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 42058 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 42059 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1639, + "step": 42060 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 42061 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 42062 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 42063 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 42064 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 42065 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 42066 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 42067 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 42068 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 42069 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 42070 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 42071 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 42072 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 42073 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 42074 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 42075 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 42076 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 42077 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 42078 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 42079 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.0903, + "step": 42080 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1517, + "step": 42081 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 42082 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 42083 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 42084 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 42085 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.0843, + "step": 42086 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 42087 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 42088 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1525, + "step": 42089 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 42090 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 42091 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 42092 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 42093 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 42094 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 42095 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 42096 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 42097 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.076, + "step": 42098 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 42099 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 42100 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 42101 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 42102 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 42103 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 42104 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 42105 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 42106 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 42107 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 42108 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 42109 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 42110 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 42111 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1603, + "step": 42112 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 42113 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 42114 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 42115 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 42116 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 42117 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 42118 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 42119 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 42120 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 42121 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 42122 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 42123 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 42124 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 42125 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 42126 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 42127 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.0943, + "step": 42128 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 42129 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 42130 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 42131 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 42132 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 42133 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 42134 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.0936, + "step": 42135 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 42136 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 42137 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 42138 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 42139 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 42140 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 42141 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 42142 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 42143 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 42144 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 42145 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 42146 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 42147 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 42148 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 42149 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.0869, + "step": 42150 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 42151 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 42152 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 42153 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 42154 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 42155 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 42156 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 42157 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 42158 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 42159 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 42160 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1649, + "step": 42161 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 42162 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 42163 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 42164 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 42165 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 42166 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 42167 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 42168 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 42169 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 42170 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 42171 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 42172 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.0877, + "step": 42173 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 42174 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 42175 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 42176 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 42177 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 42178 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 42179 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.0858, + "step": 42180 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 42181 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 42182 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 42183 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 42184 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 42185 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.091, + "step": 42186 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 42187 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 42188 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 42189 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.0918, + "step": 42190 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 42191 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1557, + "step": 42192 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 42193 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 42194 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 42195 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 42196 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 42197 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 42198 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 42199 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 42200 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 42201 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 42202 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 42203 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 42204 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 42205 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 42206 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 42207 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 42208 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.0859, + "step": 42209 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 42210 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 42211 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 42212 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 42213 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 42214 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 42215 + }, + { + "epoch": 2.56, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 42216 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 42217 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.16, + "step": 42218 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 42219 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 42220 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 42221 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 42222 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 42223 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 42224 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.0967, + "step": 42225 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 42226 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.0881, + "step": 42227 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 42228 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 42229 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 42230 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 42231 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 42232 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 42233 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 42234 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 42235 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 42236 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1588, + "step": 42237 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 42238 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 42239 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 42240 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1639, + "step": 42241 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 42242 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 42243 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 42244 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 42245 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 42246 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.0871, + "step": 42247 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 42248 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 42249 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 42250 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 42251 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 42252 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 42253 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 42254 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 42255 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 42256 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 42257 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 42258 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 42259 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1517, + "step": 42260 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 42261 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 42262 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1668, + "step": 42263 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 42264 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 42265 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.085, + "step": 42266 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 42267 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 42268 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 42269 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 42270 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 42271 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 42272 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 42273 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1478, + "step": 42274 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 42275 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 42276 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 42277 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 42278 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 42279 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 42280 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 42281 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 42282 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 42283 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 42284 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.0871, + "step": 42285 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 42286 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 42287 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 42288 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.088, + "step": 42289 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 42290 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 42291 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 42292 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 42293 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.0864, + "step": 42294 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1429, + "step": 42295 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 42296 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 42297 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 42298 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.0899, + "step": 42299 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 42300 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 42301 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 42302 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 42303 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 42304 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 42305 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 42306 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 42307 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 42308 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 42309 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 42310 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 42311 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 42312 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 42313 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 42314 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1638, + "step": 42315 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 42316 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 42317 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.0868, + "step": 42318 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 42319 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 42320 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 42321 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.0879, + "step": 42322 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 42323 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 42324 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.153, + "step": 42325 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 42326 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 42327 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 42328 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 42329 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 42330 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 42331 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.0915, + "step": 42332 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 42333 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 42334 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 42335 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 42336 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 42337 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 42338 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 42339 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.0898, + "step": 42340 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 42341 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 42342 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 42343 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.0829, + "step": 42344 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 42345 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 42346 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 42347 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 42348 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 42349 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 42350 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 42351 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 42352 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 42353 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 42354 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 42355 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1746, + "step": 42356 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 42357 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 42358 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 42359 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 42360 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 42361 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 42362 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 42363 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 42364 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 42365 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 42366 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 42367 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 42368 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1683, + "step": 42369 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1535, + "step": 42370 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 42371 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 42372 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 42373 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 42374 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 42375 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 42376 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 42377 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 42378 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 42379 + }, + { + "epoch": 2.57, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 42380 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1589, + "step": 42381 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1451, + "step": 42382 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 42383 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 42384 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 42385 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 42386 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 42387 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1526, + "step": 42388 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 42389 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 42390 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 42391 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 42392 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 42393 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 42394 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 42395 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 42396 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 42397 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 42398 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 42399 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 42400 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 42401 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 42402 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1781, + "step": 42403 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 42404 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 42405 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.0875, + "step": 42406 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 42407 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 42408 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 42409 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 42410 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 42411 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 42412 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 42413 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 42414 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 42415 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 42416 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1505, + "step": 42417 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 42418 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1442, + "step": 42419 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 42420 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 42421 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 42422 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 42423 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.0898, + "step": 42424 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 42425 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 42426 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 42427 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 42428 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 42429 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 42430 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1507, + "step": 42431 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1577, + "step": 42432 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 42433 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 42434 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 42435 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 42436 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 42437 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 42438 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 42439 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 42440 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 42441 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 42442 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 42443 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 42444 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1451, + "step": 42445 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 42446 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 42447 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 42448 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 42449 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 42450 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 42451 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1607, + "step": 42452 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 42453 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 42454 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 42455 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 42456 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 42457 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 42458 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 42459 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 42460 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 42461 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 42462 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 42463 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 42464 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 42465 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 42466 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 42467 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 42468 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 42469 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 42470 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 42471 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 42472 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 42473 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 42474 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 42475 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 42476 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 42477 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 42478 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 42479 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 42480 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1456, + "step": 42481 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.0758, + "step": 42482 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 42483 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 42484 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 42485 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 42486 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 42487 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 42488 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 42489 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 42490 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.0824, + "step": 42491 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 42492 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 42493 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 42494 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 42495 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 42496 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 42497 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 42498 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 42499 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.0916, + "step": 42500 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 42501 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 42502 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 42503 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 42504 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 42505 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 42506 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 42507 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 42508 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1442, + "step": 42509 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 42510 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 42511 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 42512 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 42513 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 42514 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 42515 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 42516 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 42517 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 42518 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.0868, + "step": 42519 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 42520 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 42521 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.0952, + "step": 42522 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 42523 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 42524 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 42525 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 42526 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1782, + "step": 42527 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 42528 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 42529 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 42530 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 42531 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 42532 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 42533 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 42534 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 42535 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 42536 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1459, + "step": 42537 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 42538 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 42539 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 42540 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 42541 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.0906, + "step": 42542 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 42543 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 42544 + }, + { + "epoch": 2.58, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 42545 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 42546 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 42547 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 42548 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 42549 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 42550 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 42551 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 42552 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 42553 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 42554 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 42555 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 42556 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 42557 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.0918, + "step": 42558 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 42559 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 42560 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 42561 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 42562 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.0755, + "step": 42563 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 42564 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 42565 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 42566 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1608, + "step": 42567 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 42568 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 42569 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 42570 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 42571 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 42572 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 42573 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 42574 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 42575 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 42576 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 42577 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 42578 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 42579 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 42580 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 42581 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 42582 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.091, + "step": 42583 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 42584 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 42585 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.0865, + "step": 42586 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 42587 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 42588 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 42589 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 42590 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 42591 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.0878, + "step": 42592 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 42593 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 42594 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 42595 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 42596 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 42597 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 42598 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 42599 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 42600 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 42601 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 42602 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 42603 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1549, + "step": 42604 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 42605 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 42606 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 42607 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 42608 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 42609 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 42610 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 42611 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 42612 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 42613 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 42614 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 42615 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 42616 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 42617 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 42618 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 42619 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1674, + "step": 42620 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1527, + "step": 42621 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 42622 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 42623 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 42624 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 42625 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 42626 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 42627 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 42628 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.0901, + "step": 42629 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 42630 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 42631 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 42632 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 42633 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.0916, + "step": 42634 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 42635 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 42636 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 42637 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 42638 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 42639 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 42640 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 42641 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 42642 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 42643 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 42644 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 42645 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 42646 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1453, + "step": 42647 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 42648 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 42649 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 42650 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 42651 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 42652 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1495, + "step": 42653 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1524, + "step": 42654 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 42655 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 42656 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 42657 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 42658 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1623, + "step": 42659 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.0883, + "step": 42660 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 42661 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 42662 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.158, + "step": 42663 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 42664 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 42665 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 42666 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 42667 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 42668 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 42669 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 42670 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 42671 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1498, + "step": 42672 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 42673 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 42674 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 42675 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 42676 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 42677 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 42678 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 42679 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 42680 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 42681 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 42682 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 42683 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 42684 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 42685 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 42686 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1774, + "step": 42687 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 42688 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1429, + "step": 42689 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 42690 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 42691 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 42692 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 42693 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 42694 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 42695 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.0763, + "step": 42696 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.0903, + "step": 42697 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 42698 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1704, + "step": 42699 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 42700 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 42701 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 42702 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 42703 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 42704 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 42705 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 42706 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 42707 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 42708 + }, + { + "epoch": 2.59, + "learning_rate": 2e-05, + "loss": 0.1596, + "step": 42709 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 42710 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 42711 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 42712 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 42713 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 42714 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 42715 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 42716 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.0876, + "step": 42717 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 42718 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 42719 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 42720 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 42721 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 42722 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 42723 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 42724 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 42725 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 42726 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 42727 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 42728 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.0803, + "step": 42729 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1548, + "step": 42730 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 42731 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 42732 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 42733 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 42734 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 42735 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 42736 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 42737 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1453, + "step": 42738 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 42739 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 42740 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 42741 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 42742 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 42743 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 42744 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1601, + "step": 42745 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1488, + "step": 42746 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 42747 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 42748 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 42749 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 42750 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 42751 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 42752 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 42753 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 42754 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 42755 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 42756 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 42757 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 42758 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 42759 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 42760 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 42761 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 42762 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 42763 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 42764 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 42765 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 42766 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 42767 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 42768 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 42769 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 42770 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.0836, + "step": 42771 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 42772 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 42773 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 42774 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 42775 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 42776 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 42777 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 42778 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 42779 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 42780 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 42781 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 42782 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 42783 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 42784 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 42785 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.0878, + "step": 42786 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 42787 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 42788 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 42789 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 42790 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 42791 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 42792 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.0967, + "step": 42793 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 42794 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 42795 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 42796 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.0943, + "step": 42797 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 42798 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 42799 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 42800 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 42801 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 42802 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 42803 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 42804 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.082, + "step": 42805 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.088, + "step": 42806 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 42807 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 42808 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 42809 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 42810 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 42811 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 42812 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 42813 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.0967, + "step": 42814 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 42815 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 42816 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 42817 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 42818 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 42819 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 42820 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 42821 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 42822 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 42823 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 42824 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 42825 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 42826 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 42827 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 42828 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 42829 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.0879, + "step": 42830 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 42831 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 42832 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 42833 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 42834 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 42835 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 42836 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 42837 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 42838 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 42839 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 42840 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 42841 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 42842 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 42843 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 42844 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 42845 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 42846 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 42847 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 42848 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 42849 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 42850 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 42851 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.0876, + "step": 42852 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 42853 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 42854 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 42855 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 42856 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 42857 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 42858 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 42859 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 42860 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 42861 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 42862 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 42863 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 42864 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 42865 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1502, + "step": 42866 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 42867 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 42868 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.0827, + "step": 42869 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 42870 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 42871 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 42872 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 42873 + }, + { + "epoch": 2.6, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 42874 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 42875 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 42876 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 42877 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 42878 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1614, + "step": 42879 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 42880 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 42881 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 42882 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 42883 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 42884 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 42885 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 42886 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 42887 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 42888 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 42889 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 42890 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 42891 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 42892 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 42893 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 42894 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 42895 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 42896 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 42897 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 42898 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 42899 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 42900 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 42901 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 42902 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 42903 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 42904 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 42905 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1546, + "step": 42906 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 42907 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1517, + "step": 42908 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 42909 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 42910 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 42911 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 42912 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 42913 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 42914 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.091, + "step": 42915 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.147, + "step": 42916 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 42917 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.0818, + "step": 42918 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 42919 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 42920 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 42921 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 42922 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 42923 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.0889, + "step": 42924 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 42925 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 42926 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 42927 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 42928 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 42929 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 42930 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 42931 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 42932 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 42933 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 42934 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 42935 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 42936 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 42937 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 42938 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 42939 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 42940 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 42941 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 42942 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 42943 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 42944 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 42945 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 42946 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 42947 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.0767, + "step": 42948 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 42949 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 42950 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 42951 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 42952 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 42953 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 42954 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 42955 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 42956 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 42957 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1533, + "step": 42958 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 42959 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 42960 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 42961 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.0875, + "step": 42962 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 42963 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 42964 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 42965 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 42966 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1593, + "step": 42967 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 42968 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 42969 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 42970 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 42971 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 42972 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 42973 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 42974 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 42975 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 42976 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 42977 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 42978 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 42979 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 42980 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 42981 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 42982 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 42983 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 42984 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 42985 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 42986 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 42987 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.0943, + "step": 42988 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 42989 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 42990 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 42991 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 42992 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 42993 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 42994 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 42995 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 42996 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 42997 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 42998 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.0755, + "step": 42999 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 43000 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 43001 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 43002 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 43003 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 43004 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 43005 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.0828, + "step": 43006 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 43007 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 43008 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 43009 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 43010 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 43011 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 43012 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 43013 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1561, + "step": 43014 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 43015 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 43016 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 43017 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 43018 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 43019 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 43020 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 43021 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 43022 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 43023 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 43024 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.0713, + "step": 43025 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 43026 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 43027 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 43028 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 43029 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 43030 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 43031 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 43032 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 43033 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 43034 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 43035 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.0909, + "step": 43036 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.0935, + "step": 43037 + }, + { + "epoch": 2.61, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 43038 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 43039 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 43040 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 43041 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 43042 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 43043 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1519, + "step": 43044 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 43045 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.0856, + "step": 43046 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 43047 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 43048 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.071, + "step": 43049 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 43050 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 43051 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 43052 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 43053 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 43054 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 43055 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 43056 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 43057 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 43058 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 43059 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 43060 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 43061 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 43062 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 43063 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 43064 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 43065 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 43066 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 43067 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 43068 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.08, + "step": 43069 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 43070 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 43071 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 43072 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 43073 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 43074 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 43075 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 43076 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 43077 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 43078 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 43079 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1618, + "step": 43080 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 43081 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 43082 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 43083 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 43084 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 43085 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 43086 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 43087 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 43088 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 43089 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 43090 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 43091 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 43092 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 43093 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 43094 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 43095 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 43096 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 43097 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 43098 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.0899, + "step": 43099 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 43100 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 43101 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 43102 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.0774, + "step": 43103 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 43104 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 43105 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1511, + "step": 43106 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.089, + "step": 43107 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 43108 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 43109 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 43110 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 43111 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 43112 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 43113 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 43114 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 43115 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 43116 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 43117 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 43118 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 43119 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 43120 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 43121 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 43122 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 43123 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 43124 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1637, + "step": 43125 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 43126 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 43127 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.0847, + "step": 43128 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 43129 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 43130 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 43131 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 43132 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 43133 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 43134 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.16, + "step": 43135 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 43136 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 43137 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 43138 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 43139 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 43140 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 43141 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 43142 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 43143 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 43144 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 43145 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 43146 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 43147 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 43148 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 43149 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 43150 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 43151 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 43152 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 43153 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 43154 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 43155 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 43156 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 43157 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 43158 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1447, + "step": 43159 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 43160 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 43161 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 43162 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 43163 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 43164 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 43165 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 43166 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.0871, + "step": 43167 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 43168 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 43169 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 43170 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.0909, + "step": 43171 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 43172 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 43173 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.0891, + "step": 43174 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 43175 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.091, + "step": 43176 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 43177 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 43178 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 43179 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.091, + "step": 43180 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 43181 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 43182 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 43183 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 43184 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 43185 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 43186 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 43187 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 43188 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 43189 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 43190 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 43191 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 43192 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 43193 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 43194 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 43195 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 43196 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 43197 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 43198 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1513, + "step": 43199 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 43200 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 43201 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 43202 + }, + { + "epoch": 2.62, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 43203 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 43204 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 43205 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 43206 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 43207 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 43208 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 43209 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 43210 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 43211 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 43212 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1574, + "step": 43213 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.144, + "step": 43214 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.0839, + "step": 43215 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 43216 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 43217 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1451, + "step": 43218 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 43219 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 43220 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 43221 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 43222 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 43223 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 43224 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 43225 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 43226 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1672, + "step": 43227 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 43228 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 43229 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 43230 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1922, + "step": 43231 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 43232 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 43233 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 43234 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 43235 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 43236 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 43237 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 43238 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 43239 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.0967, + "step": 43240 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 43241 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 43242 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 43243 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 43244 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 43245 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 43246 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 43247 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 43248 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 43249 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 43250 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 43251 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 43252 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 43253 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 43254 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 43255 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 43256 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 43257 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 43258 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 43259 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 43260 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 43261 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 43262 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.0707, + "step": 43263 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 43264 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 43265 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 43266 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 43267 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.0869, + "step": 43268 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 43269 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.0871, + "step": 43270 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 43271 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 43272 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 43273 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 43274 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 43275 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 43276 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 43277 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 43278 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 43279 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 43280 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 43281 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 43282 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 43283 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 43284 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 43285 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1696, + "step": 43286 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 43287 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 43288 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 43289 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 43290 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 43291 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 43292 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 43293 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 43294 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 43295 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 43296 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 43297 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 43298 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 43299 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 43300 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 43301 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 43302 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 43303 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 43304 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1667, + "step": 43305 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 43306 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 43307 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 43308 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 43309 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 43310 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 43311 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 43312 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.0952, + "step": 43313 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 43314 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.0914, + "step": 43315 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 43316 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.0932, + "step": 43317 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 43318 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 43319 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 43320 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 43321 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 43322 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 43323 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1429, + "step": 43324 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.0932, + "step": 43325 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 43326 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.0831, + "step": 43327 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 43328 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 43329 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 43330 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 43331 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 43332 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 43333 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 43334 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1541, + "step": 43335 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.0801, + "step": 43336 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 43337 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 43338 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 43339 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 43340 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 43341 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 43342 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 43343 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 43344 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 43345 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 43346 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 43347 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 43348 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 43349 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 43350 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 43351 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 43352 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.158, + "step": 43353 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 43354 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 43355 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 43356 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 43357 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 43358 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.091, + "step": 43359 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 43360 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 43361 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 43362 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 43363 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 43364 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 43365 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 43366 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 43367 + }, + { + "epoch": 2.63, + "learning_rate": 2e-05, + "loss": 0.1634, + "step": 43368 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 43369 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 43370 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 43371 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 43372 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 43373 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 43374 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 43375 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 43376 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 43377 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1635, + "step": 43378 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 43379 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 43380 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 43381 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.0899, + "step": 43382 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 43383 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1497, + "step": 43384 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 43385 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 43386 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 43387 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.0764, + "step": 43388 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 43389 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 43390 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 43391 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 43392 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 43393 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.0791, + "step": 43394 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 43395 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.08, + "step": 43396 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1451, + "step": 43397 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 43398 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 43399 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 43400 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 43401 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 43402 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 43403 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 43404 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 43405 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 43406 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 43407 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1694, + "step": 43408 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 43409 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 43410 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 43411 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.0765, + "step": 43412 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 43413 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 43414 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 43415 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 43416 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 43417 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 43418 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 43419 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 43420 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 43421 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 43422 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 43423 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 43424 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 43425 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 43426 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.0883, + "step": 43427 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 43428 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 43429 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 43430 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 43431 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 43432 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 43433 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 43434 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 43435 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 43436 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 43437 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 43438 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 43439 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 43440 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 43441 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 43442 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 43443 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 43444 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 43445 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 43446 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 43447 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 43448 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 43449 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 43450 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 43451 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 43452 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.0895, + "step": 43453 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 43454 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 43455 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 43456 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 43457 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 43458 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 43459 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 43460 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 43461 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 43462 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 43463 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 43464 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 43465 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 43466 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 43467 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.0878, + "step": 43468 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 43469 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 43470 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 43471 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 43472 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 43473 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 43474 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 43475 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 43476 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 43477 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 43478 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 43479 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 43480 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 43481 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 43482 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 43483 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 43484 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 43485 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 43486 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 43487 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 43488 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 43489 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 43490 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 43491 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.089, + "step": 43492 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 43493 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 43494 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 43495 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 43496 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 43497 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 43498 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 43499 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1756, + "step": 43500 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 43501 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1632, + "step": 43502 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 43503 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 43504 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1648, + "step": 43505 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 43506 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 43507 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 43508 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 43509 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 43510 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 43511 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 43512 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 43513 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 43514 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 43515 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.0791, + "step": 43516 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 43517 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 43518 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 43519 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 43520 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1659, + "step": 43521 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 43522 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 43523 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 43524 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 43525 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.0903, + "step": 43526 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1505, + "step": 43527 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 43528 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 43529 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 43530 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 43531 + }, + { + "epoch": 2.64, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 43532 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 43533 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 43534 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 43535 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 43536 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 43537 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 43538 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1646, + "step": 43539 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 43540 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 43541 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 43542 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 43543 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 43544 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 43545 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 43546 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 43547 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 43548 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 43549 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 43550 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 43551 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1515, + "step": 43552 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 43553 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 43554 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 43555 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 43556 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 43557 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 43558 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 43559 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 43560 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 43561 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 43562 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 43563 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 43564 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 43565 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 43566 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 43567 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 43568 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 43569 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1549, + "step": 43570 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1497, + "step": 43571 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 43572 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 43573 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 43574 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 43575 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 43576 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1577, + "step": 43577 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 43578 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 43579 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.0774, + "step": 43580 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 43581 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 43582 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 43583 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 43584 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.0892, + "step": 43585 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 43586 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 43587 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 43588 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 43589 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 43590 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 43591 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 43592 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 43593 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 43594 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 43595 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 43596 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 43597 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 43598 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1429, + "step": 43599 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1524, + "step": 43600 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 43601 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 43602 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 43603 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1553, + "step": 43604 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 43605 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 43606 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 43607 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1495, + "step": 43608 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 43609 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 43610 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 43611 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 43612 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 43613 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 43614 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 43615 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 43616 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1567, + "step": 43617 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 43618 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 43619 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.0889, + "step": 43620 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 43621 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 43622 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 43623 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 43624 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 43625 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 43626 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 43627 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 43628 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 43629 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 43630 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 43631 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 43632 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1625, + "step": 43633 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 43634 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 43635 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 43636 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 43637 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.153, + "step": 43638 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 43639 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 43640 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 43641 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 43642 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 43643 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 43644 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 43645 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 43646 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 43647 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 43648 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 43649 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 43650 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 43651 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 43652 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 43653 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 43654 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 43655 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 43656 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 43657 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 43658 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 43659 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 43660 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 43661 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1769, + "step": 43662 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 43663 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.0847, + "step": 43664 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 43665 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 43666 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 43667 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 43668 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 43669 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 43670 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 43671 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 43672 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 43673 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 43674 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 43675 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 43676 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 43677 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.0871, + "step": 43678 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 43679 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 43680 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 43681 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 43682 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 43683 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 43684 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 43685 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 43686 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 43687 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 43688 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 43689 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 43690 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 43691 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 43692 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 43693 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.0847, + "step": 43694 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1728, + "step": 43695 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 43696 + }, + { + "epoch": 2.65, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 43697 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 43698 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1847, + "step": 43699 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 43700 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 43701 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 43702 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 43703 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 43704 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 43705 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 43706 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 43707 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 43708 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 43709 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 43710 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.0901, + "step": 43711 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.0918, + "step": 43712 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 43713 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 43714 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 43715 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 43716 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.0849, + "step": 43717 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 43718 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 43719 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 43720 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 43721 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 43722 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 43723 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 43724 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 43725 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 43726 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 43727 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 43728 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 43729 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 43730 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 43731 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 43732 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 43733 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 43734 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1566, + "step": 43735 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 43736 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 43737 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 43738 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1538, + "step": 43739 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 43740 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 43741 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 43742 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 43743 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 43744 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 43745 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 43746 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 43747 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 43748 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 43749 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 43750 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1586, + "step": 43751 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 43752 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 43753 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 43754 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 43755 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 43756 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 43757 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 43758 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 43759 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 43760 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 43761 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 43762 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 43763 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 43764 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 43765 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 43766 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 43767 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.154, + "step": 43768 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 43769 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 43770 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 43771 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 43772 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 43773 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.0788, + "step": 43774 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 43775 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1521, + "step": 43776 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 43777 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 43778 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 43779 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 43780 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 43781 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 43782 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1721, + "step": 43783 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 43784 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 43785 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.0897, + "step": 43786 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1744, + "step": 43787 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 43788 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 43789 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 43790 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 43791 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 43792 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 43793 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 43794 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 43795 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 43796 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 43797 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 43798 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 43799 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 43800 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 43801 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 43802 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 43803 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 43804 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 43805 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 43806 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 43807 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 43808 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 43809 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.0833, + "step": 43810 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 43811 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 43812 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 43813 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 43814 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 43815 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 43816 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 43817 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1476, + "step": 43818 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 43819 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 43820 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 43821 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 43822 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 43823 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 43824 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 43825 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 43826 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 43827 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 43828 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 43829 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 43830 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 43831 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.0854, + "step": 43832 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 43833 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 43834 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 43835 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 43836 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 43837 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 43838 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 43839 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 43840 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 43841 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 43842 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.09, + "step": 43843 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 43844 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 43845 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.0868, + "step": 43846 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 43847 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 43848 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 43849 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 43850 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.074, + "step": 43851 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 43852 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 43853 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.0806, + "step": 43854 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 43855 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.0872, + "step": 43856 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 43857 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 43858 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 43859 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 43860 + }, + { + "epoch": 2.66, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 43861 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 43862 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 43863 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 43864 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 43865 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 43866 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 43867 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 43868 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 43869 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 43870 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.0878, + "step": 43871 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 43872 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 43873 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 43874 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 43875 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 43876 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 43877 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 43878 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.0842, + "step": 43879 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 43880 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 43881 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.154, + "step": 43882 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 43883 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 43884 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 43885 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1447, + "step": 43886 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 43887 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1576, + "step": 43888 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 43889 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 43890 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 43891 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 43892 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1532, + "step": 43893 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 43894 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 43895 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 43896 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 43897 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 43898 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1499, + "step": 43899 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 43900 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 43901 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 43902 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 43903 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 43904 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 43905 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 43906 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 43907 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 43908 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.0932, + "step": 43909 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 43910 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 43911 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 43912 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 43913 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 43914 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 43915 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 43916 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1612, + "step": 43917 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 43918 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1513, + "step": 43919 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1593, + "step": 43920 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 43921 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 43922 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 43923 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 43924 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 43925 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.0914, + "step": 43926 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 43927 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 43928 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 43929 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 43930 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 43931 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 43932 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 43933 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 43934 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 43935 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.078, + "step": 43936 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 43937 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 43938 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 43939 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 43940 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 43941 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 43942 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 43943 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 43944 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 43945 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 43946 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1505, + "step": 43947 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 43948 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1468, + "step": 43949 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 43950 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1478, + "step": 43951 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 43952 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 43953 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1453, + "step": 43954 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 43955 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 43956 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 43957 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.0842, + "step": 43958 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.148, + "step": 43959 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 43960 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 43961 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1675, + "step": 43962 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 43963 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 43964 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.0894, + "step": 43965 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 43966 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 43967 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 43968 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 43969 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 43970 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 43971 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 43972 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 43973 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 43974 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 43975 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1795, + "step": 43976 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 43977 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 43978 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 43979 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 43980 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 43981 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 43982 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 43983 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 43984 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 43985 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 43986 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 43987 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 43988 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 43989 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 43990 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.0766, + "step": 43991 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 43992 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1468, + "step": 43993 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 43994 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1567, + "step": 43995 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 43996 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 43997 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1528, + "step": 43998 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 43999 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 44000 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 44001 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 44002 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 44003 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 44004 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.0889, + "step": 44005 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 44006 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 44007 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 44008 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 44009 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 44010 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 44011 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 44012 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 44013 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 44014 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 44015 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 44016 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 44017 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 44018 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 44019 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 44020 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 44021 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 44022 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 44023 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 44024 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 44025 + }, + { + "epoch": 2.67, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 44026 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 44027 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 44028 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 44029 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 44030 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 44031 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 44032 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 44033 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 44034 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 44035 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 44036 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1599, + "step": 44037 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 44038 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 44039 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 44040 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 44041 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 44042 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 44043 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 44044 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1656, + "step": 44045 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 44046 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 44047 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 44048 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 44049 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.0892, + "step": 44050 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 44051 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 44052 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 44053 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 44054 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 44055 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 44056 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 44057 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 44058 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 44059 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1689, + "step": 44060 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 44061 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 44062 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 44063 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 44064 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 44065 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 44066 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 44067 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 44068 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 44069 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 44070 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.0858, + "step": 44071 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 44072 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 44073 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 44074 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 44075 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 44076 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.0811, + "step": 44077 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 44078 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 44079 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.0852, + "step": 44080 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 44081 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 44082 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 44083 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 44084 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 44085 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1443, + "step": 44086 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 44087 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 44088 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 44089 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 44090 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.0952, + "step": 44091 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 44092 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 44093 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1696, + "step": 44094 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 44095 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 44096 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 44097 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 44098 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1495, + "step": 44099 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 44100 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 44101 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 44102 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.0909, + "step": 44103 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 44104 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.091, + "step": 44105 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 44106 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 44107 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 44108 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 44109 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 44110 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 44111 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 44112 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 44113 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 44114 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 44115 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 44116 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 44117 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 44118 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 44119 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 44120 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 44121 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 44122 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 44123 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 44124 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 44125 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 44126 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 44127 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1784, + "step": 44128 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 44129 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 44130 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 44131 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 44132 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 44133 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 44134 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 44135 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 44136 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 44137 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 44138 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 44139 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 44140 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 44141 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 44142 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.0781, + "step": 44143 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 44144 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 44145 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1606, + "step": 44146 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 44147 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 44148 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 44149 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.0792, + "step": 44150 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 44151 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 44152 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 44153 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 44154 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 44155 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 44156 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 44157 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1533, + "step": 44158 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 44159 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 44160 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 44161 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 44162 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.0804, + "step": 44163 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 44164 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 44165 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 44166 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 44167 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 44168 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 44169 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 44170 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1554, + "step": 44171 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 44172 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 44173 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 44174 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1469, + "step": 44175 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 44176 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 44177 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 44178 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 44179 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 44180 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 44181 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1521, + "step": 44182 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 44183 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 44184 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 44185 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 44186 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 44187 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 44188 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 44189 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 44190 + }, + { + "epoch": 2.68, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 44191 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 44192 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 44193 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 44194 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1518, + "step": 44195 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 44196 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.0936, + "step": 44197 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 44198 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 44199 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 44200 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 44201 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 44202 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 44203 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 44204 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 44205 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 44206 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 44207 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 44208 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 44209 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 44210 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1521, + "step": 44211 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 44212 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 44213 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 44214 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 44215 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.0851, + "step": 44216 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 44217 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 44218 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 44219 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 44220 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 44221 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 44222 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 44223 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 44224 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 44225 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 44226 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 44227 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 44228 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 44229 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 44230 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 44231 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.0883, + "step": 44232 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 44233 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 44234 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 44235 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.0765, + "step": 44236 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 44237 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 44238 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 44239 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 44240 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 44241 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 44242 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 44243 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 44244 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 44245 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.0761, + "step": 44246 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 44247 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 44248 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 44249 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 44250 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 44251 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 44252 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.091, + "step": 44253 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.146, + "step": 44254 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 44255 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 44256 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1524, + "step": 44257 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 44258 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 44259 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 44260 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 44261 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 44262 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 44263 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 44264 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.0876, + "step": 44265 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 44266 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.0865, + "step": 44267 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 44268 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 44269 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.0857, + "step": 44270 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 44271 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 44272 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 44273 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 44274 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 44275 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 44276 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 44277 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 44278 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 44279 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 44280 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 44281 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 44282 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.0851, + "step": 44283 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 44284 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 44285 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 44286 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 44287 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 44288 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 44289 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 44290 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 44291 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 44292 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 44293 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 44294 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 44295 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1595, + "step": 44296 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.0851, + "step": 44297 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 44298 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 44299 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 44300 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 44301 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.0852, + "step": 44302 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 44303 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 44304 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.0892, + "step": 44305 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 44306 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 44307 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 44308 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 44309 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 44310 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 44311 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1856, + "step": 44312 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 44313 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 44314 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 44315 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 44316 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 44317 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 44318 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 44319 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1503, + "step": 44320 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 44321 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 44322 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 44323 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 44324 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 44325 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 44326 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 44327 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.09, + "step": 44328 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 44329 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 44330 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 44331 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 44332 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 44333 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 44334 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 44335 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 44336 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 44337 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 44338 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 44339 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 44340 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 44341 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 44342 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.0917, + "step": 44343 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 44344 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 44345 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 44346 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 44347 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 44348 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 44349 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 44350 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 44351 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 44352 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.0854, + "step": 44353 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 44354 + }, + { + "epoch": 2.69, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 44355 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 44356 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 44357 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 44358 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 44359 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 44360 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.0914, + "step": 44361 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 44362 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 44363 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 44364 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 44365 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 44366 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 44367 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 44368 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 44369 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 44370 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 44371 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 44372 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 44373 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 44374 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 44375 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.0796, + "step": 44376 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 44377 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 44378 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 44379 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 44380 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.0894, + "step": 44381 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 44382 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 44383 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 44384 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 44385 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.0929, + "step": 44386 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 44387 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 44388 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 44389 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 44390 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 44391 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 44392 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 44393 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 44394 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 44395 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 44396 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 44397 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 44398 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 44399 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 44400 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 44401 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 44402 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 44403 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 44404 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 44405 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 44406 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 44407 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 44408 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1638, + "step": 44409 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 44410 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 44411 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 44412 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 44413 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 44414 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 44415 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 44416 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 44417 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.16, + "step": 44418 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.0967, + "step": 44419 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 44420 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 44421 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 44422 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 44423 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 44424 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 44425 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 44426 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 44427 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 44428 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1469, + "step": 44429 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 44430 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 44431 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 44432 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1699, + "step": 44433 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 44434 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 44435 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 44436 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 44437 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 44438 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 44439 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 44440 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 44441 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 44442 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1716, + "step": 44443 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 44444 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 44445 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 44446 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.162, + "step": 44447 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 44448 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 44449 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 44450 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 44451 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 44452 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 44453 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 44454 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 44455 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 44456 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 44457 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 44458 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 44459 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 44460 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 44461 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.0943, + "step": 44462 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 44463 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 44464 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 44465 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 44466 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 44467 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 44468 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 44469 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 44470 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.0851, + "step": 44471 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 44472 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 44473 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 44474 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 44475 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 44476 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 44477 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 44478 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 44479 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 44480 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 44481 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 44482 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 44483 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 44484 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 44485 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 44486 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 44487 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 44488 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 44489 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 44490 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 44491 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 44492 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 44493 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 44494 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 44495 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 44496 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 44497 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 44498 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 44499 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 44500 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.2003, + "step": 44501 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 44502 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 44503 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 44504 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 44505 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 44506 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 44507 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 44508 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 44509 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 44510 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 44511 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 44512 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 44513 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1611, + "step": 44514 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 44515 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.0834, + "step": 44516 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 44517 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 44518 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 44519 + }, + { + "epoch": 2.7, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 44520 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 44521 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 44522 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 44523 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 44524 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 44525 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.0838, + "step": 44526 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 44527 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 44528 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 44529 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1507, + "step": 44530 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 44531 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 44532 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 44533 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 44534 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 44535 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 44536 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 44537 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 44538 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.0892, + "step": 44539 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 44540 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 44541 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 44542 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 44543 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 44544 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 44545 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 44546 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 44547 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 44548 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 44549 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 44550 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 44551 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 44552 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 44553 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1591, + "step": 44554 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 44555 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 44556 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.08, + "step": 44557 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 44558 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.155, + "step": 44559 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 44560 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 44561 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 44562 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 44563 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 44564 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1681, + "step": 44565 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 44566 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 44567 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 44568 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 44569 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 44570 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 44571 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 44572 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 44573 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 44574 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 44575 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 44576 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 44577 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 44578 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 44579 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 44580 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 44581 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 44582 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.0903, + "step": 44583 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 44584 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 44585 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 44586 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 44587 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 44588 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 44589 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 44590 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 44591 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 44592 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 44593 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 44594 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 44595 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 44596 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 44597 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 44598 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 44599 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 44600 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 44601 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 44602 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 44603 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 44604 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 44605 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 44606 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 44607 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 44608 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 44609 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 44610 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 44611 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1601, + "step": 44612 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 44613 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 44614 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 44615 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 44616 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 44617 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 44618 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1548, + "step": 44619 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 44620 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 44621 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 44622 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 44623 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.148, + "step": 44624 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 44625 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 44626 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 44627 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 44628 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 44629 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 44630 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 44631 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 44632 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 44633 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 44634 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 44635 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 44636 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 44637 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 44638 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.0868, + "step": 44639 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 44640 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1461, + "step": 44641 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.146, + "step": 44642 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 44643 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 44644 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 44645 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 44646 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 44647 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 44648 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.165, + "step": 44649 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 44650 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 44651 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 44652 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 44653 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 44654 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.0777, + "step": 44655 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 44656 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 44657 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 44658 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 44659 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 44660 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.0821, + "step": 44661 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 44662 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 44663 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1503, + "step": 44664 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 44665 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 44666 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 44667 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 44668 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 44669 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 44670 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 44671 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 44672 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 44673 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 44674 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 44675 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 44676 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 44677 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 44678 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 44679 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 44680 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 44681 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 44682 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 44683 + }, + { + "epoch": 2.71, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 44684 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 44685 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 44686 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.0803, + "step": 44687 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 44688 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 44689 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 44690 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 44691 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 44692 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 44693 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 44694 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 44695 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 44696 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 44697 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 44698 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 44699 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.0849, + "step": 44700 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 44701 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 44702 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 44703 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 44704 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1593, + "step": 44705 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 44706 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 44707 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 44708 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 44709 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 44710 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 44711 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 44712 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 44713 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 44714 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 44715 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 44716 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 44717 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 44718 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 44719 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 44720 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 44721 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 44722 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 44723 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 44724 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 44725 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 44726 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 44727 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 44728 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 44729 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 44730 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 44731 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.0807, + "step": 44732 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1608, + "step": 44733 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 44734 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 44735 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1469, + "step": 44736 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 44737 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 44738 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 44739 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 44740 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 44741 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 44742 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 44743 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 44744 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 44745 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 44746 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 44747 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 44748 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 44749 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 44750 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 44751 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 44752 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 44753 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.0849, + "step": 44754 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 44755 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 44756 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 44757 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 44758 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 44759 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 44760 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 44761 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1848, + "step": 44762 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.0862, + "step": 44763 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 44764 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.0916, + "step": 44765 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 44766 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 44767 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 44768 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 44769 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 44770 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 44771 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 44772 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.0879, + "step": 44773 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 44774 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 44775 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 44776 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 44777 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 44778 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 44779 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 44780 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 44781 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 44782 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 44783 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 44784 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 44785 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 44786 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 44787 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 44788 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 44789 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 44790 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.0725, + "step": 44791 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 44792 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 44793 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 44794 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 44795 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 44796 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 44797 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 44798 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 44799 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.0863, + "step": 44800 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 44801 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 44802 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 44803 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 44804 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1541, + "step": 44805 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 44806 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 44807 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1568, + "step": 44808 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 44809 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 44810 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1459, + "step": 44811 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 44812 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 44813 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 44814 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 44815 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 44816 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 44817 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 44818 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.152, + "step": 44819 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 44820 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.0819, + "step": 44821 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 44822 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 44823 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 44824 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 44825 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 44826 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 44827 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 44828 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 44829 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 44830 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 44831 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 44832 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 44833 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 44834 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 44835 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 44836 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 44837 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 44838 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 44839 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 44840 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 44841 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 44842 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 44843 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 44844 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 44845 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 44846 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 44847 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 44848 + }, + { + "epoch": 2.72, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 44849 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 44850 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.0929, + "step": 44851 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 44852 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 44853 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 44854 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 44855 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1749, + "step": 44856 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 44857 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 44858 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 44859 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 44860 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 44861 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 44862 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 44863 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.0899, + "step": 44864 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.146, + "step": 44865 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 44866 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 44867 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 44868 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 44869 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 44870 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 44871 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 44872 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 44873 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1765, + "step": 44874 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 44875 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.152, + "step": 44876 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 44877 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.0869, + "step": 44878 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 44879 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 44880 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.0929, + "step": 44881 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1518, + "step": 44882 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 44883 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 44884 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 44885 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 44886 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 44887 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 44888 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 44889 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 44890 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 44891 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 44892 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 44893 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 44894 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 44895 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.0783, + "step": 44896 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 44897 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 44898 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 44899 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 44900 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 44901 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 44902 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 44903 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 44904 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.0829, + "step": 44905 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 44906 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1763, + "step": 44907 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 44908 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 44909 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 44910 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 44911 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 44912 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 44913 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 44914 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 44915 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 44916 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 44917 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 44918 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 44919 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 44920 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 44921 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 44922 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 44923 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 44924 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 44925 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 44926 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 44927 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 44928 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 44929 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1447, + "step": 44930 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 44931 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 44932 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 44933 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1493, + "step": 44934 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 44935 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 44936 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 44937 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 44938 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 44939 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 44940 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 44941 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 44942 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 44943 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 44944 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 44945 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 44946 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 44947 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 44948 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 44949 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 44950 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 44951 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 44952 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 44953 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 44954 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 44955 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 44956 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 44957 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 44958 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 44959 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 44960 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 44961 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 44962 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 44963 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.0863, + "step": 44964 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 44965 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 44966 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 44967 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 44968 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 44969 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 44970 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 44971 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 44972 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 44973 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 44974 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 44975 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 44976 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 44977 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 44978 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.0739, + "step": 44979 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 44980 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 44981 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 44982 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 44983 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 44984 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1749, + "step": 44985 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 44986 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 44987 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 44988 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 44989 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 44990 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 44991 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 44992 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 44993 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 44994 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 44995 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 44996 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 44997 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 44998 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 44999 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 45000 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 45001 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1741, + "step": 45002 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 45003 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 45004 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 45005 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 45006 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.0872, + "step": 45007 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 45008 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 45009 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 45010 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.0879, + "step": 45011 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 45012 + }, + { + "epoch": 2.73, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 45013 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 45014 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 45015 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 45016 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 45017 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 45018 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 45019 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 45020 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 45021 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 45022 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 45023 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 45024 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 45025 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 45026 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.0857, + "step": 45027 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 45028 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 45029 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 45030 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 45031 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 45032 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 45033 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 45034 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 45035 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 45036 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 45037 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 45038 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 45039 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 45040 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 45041 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 45042 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 45043 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 45044 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 45045 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 45046 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 45047 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 45048 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 45049 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.076, + "step": 45050 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 45051 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 45052 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 45053 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 45054 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 45055 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 45056 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 45057 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 45058 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 45059 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 45060 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 45061 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 45062 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 45063 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 45064 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 45065 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 45066 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 45067 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 45068 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 45069 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 45070 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 45071 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 45072 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 45073 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.0902, + "step": 45074 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 45075 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 45076 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.09, + "step": 45077 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 45078 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 45079 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 45080 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 45081 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 45082 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 45083 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 45084 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 45085 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 45086 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 45087 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 45088 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 45089 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 45090 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 45091 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 45092 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 45093 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 45094 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 45095 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 45096 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 45097 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 45098 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.0869, + "step": 45099 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 45100 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 45101 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 45102 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 45103 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 45104 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 45105 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 45106 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 45107 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 45108 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1442, + "step": 45109 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 45110 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 45111 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 45112 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 45113 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 45114 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.0901, + "step": 45115 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 45116 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 45117 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 45118 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 45119 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 45120 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1739, + "step": 45121 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 45122 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 45123 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 45124 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 45125 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 45126 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 45127 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 45128 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 45129 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 45130 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 45131 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 45132 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 45133 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 45134 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 45135 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 45136 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1465, + "step": 45137 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 45138 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1637, + "step": 45139 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 45140 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.0859, + "step": 45141 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1584, + "step": 45142 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 45143 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 45144 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 45145 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 45146 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 45147 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1505, + "step": 45148 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 45149 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 45150 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 45151 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 45152 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 45153 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 45154 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 45155 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.145, + "step": 45156 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 45157 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 45158 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 45159 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 45160 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.0722, + "step": 45161 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 45162 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 45163 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 45164 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 45165 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 45166 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 45167 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 45168 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 45169 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 45170 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 45171 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 45172 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 45173 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 45174 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 45175 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 45176 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 45177 + }, + { + "epoch": 2.74, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 45178 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 45179 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 45180 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 45181 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 45182 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 45183 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 45184 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 45185 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 45186 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 45187 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 45188 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 45189 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 45190 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 45191 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 45192 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 45193 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.0865, + "step": 45194 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 45195 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.0876, + "step": 45196 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 45197 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 45198 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 45199 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 45200 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.0856, + "step": 45201 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 45202 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 45203 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 45204 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 45205 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 45206 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 45207 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 45208 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 45209 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 45210 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 45211 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 45212 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.0795, + "step": 45213 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 45214 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 45215 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 45216 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 45217 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 45218 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 45219 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 45220 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 45221 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 45222 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 45223 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 45224 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 45225 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 45226 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 45227 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 45228 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 45229 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 45230 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 45231 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 45232 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1573, + "step": 45233 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 45234 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 45235 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 45236 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.0817, + "step": 45237 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 45238 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 45239 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 45240 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 45241 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 45242 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 45243 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 45244 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1645, + "step": 45245 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 45246 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 45247 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 45248 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.0808, + "step": 45249 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 45250 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 45251 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 45252 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1444, + "step": 45253 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 45254 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 45255 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 45256 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 45257 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 45258 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 45259 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 45260 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 45261 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 45262 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 45263 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 45264 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 45265 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.0803, + "step": 45266 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 45267 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 45268 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 45269 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 45270 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 45271 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 45272 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 45273 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 45274 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 45275 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 45276 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 45277 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 45278 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.0733, + "step": 45279 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 45280 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 45281 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 45282 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 45283 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 45284 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 45285 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 45286 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 45287 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 45288 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 45289 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1462, + "step": 45290 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 45291 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.0783, + "step": 45292 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 45293 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 45294 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 45295 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 45296 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 45297 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 45298 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 45299 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 45300 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 45301 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 45302 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 45303 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 45304 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 45305 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 45306 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 45307 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 45308 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1557, + "step": 45309 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 45310 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 45311 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 45312 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 45313 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 45314 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 45315 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 45316 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 45317 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 45318 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 45319 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 45320 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 45321 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 45322 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 45323 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.0918, + "step": 45324 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 45325 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 45326 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 45327 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 45328 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 45329 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 45330 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 45331 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 45332 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 45333 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 45334 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 45335 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.0906, + "step": 45336 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 45337 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 45338 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 45339 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 45340 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 45341 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 45342 + }, + { + "epoch": 2.75, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 45343 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 45344 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 45345 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 45346 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 45347 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 45348 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 45349 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 45350 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 45351 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 45352 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 45353 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 45354 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 45355 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 45356 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 45357 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.0891, + "step": 45358 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 45359 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.088, + "step": 45360 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 45361 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 45362 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 45363 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 45364 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 45365 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 45366 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 45367 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 45368 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.0899, + "step": 45369 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 45370 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 45371 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 45372 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 45373 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 45374 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 45375 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 45376 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 45377 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 45378 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 45379 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 45380 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 45381 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 45382 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 45383 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 45384 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 45385 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 45386 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 45387 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 45388 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 45389 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 45390 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 45391 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 45392 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 45393 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 45394 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 45395 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 45396 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 45397 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 45398 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 45399 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 45400 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 45401 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.09, + "step": 45402 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 45403 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 45404 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 45405 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 45406 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 45407 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 45408 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 45409 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 45410 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 45411 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 45412 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 45413 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 45414 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 45415 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 45416 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 45417 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 45418 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.0843, + "step": 45419 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 45420 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.0901, + "step": 45421 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1557, + "step": 45422 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 45423 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 45424 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 45425 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 45426 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 45427 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 45428 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 45429 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 45430 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 45431 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 45432 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 45433 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 45434 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 45435 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 45436 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 45437 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 45438 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 45439 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 45440 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 45441 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 45442 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 45443 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 45444 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 45445 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 45446 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.0891, + "step": 45447 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 45448 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 45449 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 45450 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.0824, + "step": 45451 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.0909, + "step": 45452 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 45453 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 45454 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 45455 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 45456 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 45457 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 45458 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 45459 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 45460 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 45461 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 45462 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 45463 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 45464 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 45465 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 45466 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 45467 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 45468 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 45469 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 45470 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 45471 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 45472 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 45473 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 45474 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 45475 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 45476 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 45477 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 45478 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1462, + "step": 45479 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 45480 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.0865, + "step": 45481 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 45482 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 45483 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 45484 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 45485 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 45486 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 45487 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 45488 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 45489 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 45490 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 45491 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1459, + "step": 45492 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 45493 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 45494 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 45495 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 45496 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 45497 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 45498 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 45499 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 45500 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 45501 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 45502 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 45503 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 45504 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 45505 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 45506 + }, + { + "epoch": 2.76, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 45507 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 45508 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 45509 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 45510 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 45511 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 45512 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 45513 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 45514 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 45515 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 45516 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 45517 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 45518 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 45519 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 45520 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 45521 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 45522 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 45523 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 45524 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 45525 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.147, + "step": 45526 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 45527 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1586, + "step": 45528 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 45529 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 45530 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 45531 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 45532 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 45533 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 45534 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.146, + "step": 45535 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 45536 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 45537 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 45538 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 45539 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 45540 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 45541 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 45542 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 45543 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 45544 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 45545 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 45546 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 45547 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 45548 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 45549 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 45550 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 45551 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.082, + "step": 45552 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 45553 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.089, + "step": 45554 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 45555 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 45556 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 45557 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 45558 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 45559 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 45560 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 45561 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.0918, + "step": 45562 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 45563 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1525, + "step": 45564 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 45565 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 45566 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 45567 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 45568 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 45569 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 45570 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 45571 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1488, + "step": 45572 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 45573 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.0899, + "step": 45574 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 45575 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 45576 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 45577 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 45578 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 45579 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 45580 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 45581 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 45582 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 45583 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 45584 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 45585 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 45586 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 45587 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 45588 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 45589 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 45590 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 45591 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 45592 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 45593 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 45594 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 45595 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 45596 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 45597 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 45598 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 45599 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 45600 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 45601 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 45602 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.0911, + "step": 45603 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1566, + "step": 45604 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 45605 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 45606 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 45607 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 45608 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 45609 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1658, + "step": 45610 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 45611 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 45612 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 45613 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 45614 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 45615 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 45616 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.0869, + "step": 45617 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 45618 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 45619 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 45620 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 45621 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 45622 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 45623 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 45624 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 45625 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 45626 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 45627 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 45628 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 45629 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 45630 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 45631 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 45632 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 45633 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 45634 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 45635 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 45636 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1706, + "step": 45637 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 45638 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 45639 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 45640 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 45641 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 45642 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 45643 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 45644 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.0761, + "step": 45645 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 45646 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 45647 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 45648 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 45649 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 45650 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 45651 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 45652 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 45653 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.0899, + "step": 45654 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 45655 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 45656 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 45657 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 45658 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 45659 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.0742, + "step": 45660 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 45661 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 45662 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 45663 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 45664 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.0642, + "step": 45665 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.0795, + "step": 45666 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 45667 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 45668 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 45669 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 45670 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 45671 + }, + { + "epoch": 2.77, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 45672 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 45673 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 45674 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 45675 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 45676 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 45677 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 45678 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 45679 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 45680 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 45681 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 45682 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.0844, + "step": 45683 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.0833, + "step": 45684 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 45685 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 45686 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 45687 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 45688 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 45689 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1443, + "step": 45690 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 45691 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 45692 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 45693 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 45694 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 45695 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 45696 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 45697 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.0822, + "step": 45698 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 45699 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 45700 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 45701 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 45702 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 45703 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 45704 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 45705 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 45706 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 45707 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 45708 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 45709 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.163, + "step": 45710 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 45711 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 45712 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 45713 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 45714 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 45715 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 45716 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 45717 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 45718 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 45719 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 45720 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 45721 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 45722 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 45723 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 45724 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 45725 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1744, + "step": 45726 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 45727 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 45728 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 45729 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 45730 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 45731 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 45732 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 45733 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.0898, + "step": 45734 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 45735 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 45736 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 45737 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 45738 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 45739 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 45740 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 45741 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 45742 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 45743 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 45744 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 45745 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 45746 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 45747 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 45748 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.0914, + "step": 45749 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 45750 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1453, + "step": 45751 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 45752 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 45753 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 45754 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 45755 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 45756 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 45757 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 45758 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 45759 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 45760 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 45761 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1642, + "step": 45762 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 45763 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 45764 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 45765 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 45766 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 45767 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 45768 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 45769 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 45770 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.0874, + "step": 45771 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 45772 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1503, + "step": 45773 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 45774 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 45775 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.0875, + "step": 45776 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 45777 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 45778 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 45779 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 45780 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 45781 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.0867, + "step": 45782 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 45783 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 45784 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 45785 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 45786 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 45787 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.0879, + "step": 45788 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 45789 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 45790 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 45791 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 45792 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 45793 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 45794 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 45795 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1661, + "step": 45796 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 45797 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 45798 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1526, + "step": 45799 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 45800 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 45801 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 45802 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 45803 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.0764, + "step": 45804 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 45805 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1469, + "step": 45806 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1745, + "step": 45807 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 45808 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 45809 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 45810 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 45811 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 45812 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 45813 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 45814 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 45815 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 45816 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1635, + "step": 45817 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1456, + "step": 45818 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 45819 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 45820 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 45821 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 45822 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 45823 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 45824 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.0855, + "step": 45825 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 45826 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 45827 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 45828 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 45829 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 45830 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 45831 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 45832 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 45833 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 45834 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.0876, + "step": 45835 + }, + { + "epoch": 2.78, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 45836 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 45837 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 45838 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 45839 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 45840 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 45841 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.0807, + "step": 45842 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 45843 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 45844 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 45845 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 45846 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 45847 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 45848 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 45849 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 45850 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 45851 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 45852 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 45853 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 45854 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 45855 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 45856 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.089, + "step": 45857 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 45858 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.154, + "step": 45859 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 45860 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 45861 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 45862 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 45863 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 45864 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 45865 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 45866 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.0825, + "step": 45867 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 45868 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 45869 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 45870 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 45871 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 45872 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1453, + "step": 45873 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1524, + "step": 45874 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 45875 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 45876 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 45877 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 45878 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 45879 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.0936, + "step": 45880 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 45881 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 45882 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 45883 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 45884 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 45885 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 45886 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 45887 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 45888 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 45889 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 45890 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 45891 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 45892 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.0845, + "step": 45893 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 45894 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1554, + "step": 45895 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1542, + "step": 45896 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 45897 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 45898 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 45899 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 45900 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 45901 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.0899, + "step": 45902 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 45903 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 45904 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 45905 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 45906 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 45907 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 45908 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 45909 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 45910 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 45911 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 45912 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 45913 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 45914 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 45915 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 45916 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 45917 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 45918 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 45919 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 45920 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 45921 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.0932, + "step": 45922 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 45923 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 45924 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 45925 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1644, + "step": 45926 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 45927 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 45928 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 45929 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 45930 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 45931 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 45932 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 45933 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 45934 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 45935 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 45936 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 45937 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 45938 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1667, + "step": 45939 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 45940 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 45941 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 45942 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 45943 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 45944 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 45945 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 45946 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 45947 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 45948 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 45949 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 45950 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.145, + "step": 45951 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 45952 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 45953 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 45954 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 45955 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 45956 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1801, + "step": 45957 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 45958 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.0952, + "step": 45959 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 45960 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 45961 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 45962 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 45963 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 45964 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.0737, + "step": 45965 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 45966 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 45967 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 45968 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 45969 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 45970 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 45971 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 45972 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 45973 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 45974 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.0799, + "step": 45975 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 45976 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 45977 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 45978 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 45979 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 45980 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 45981 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 45982 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 45983 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 45984 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 45985 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 45986 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.082, + "step": 45987 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 45988 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 45989 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 45990 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 45991 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 45992 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 45993 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 45994 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 45995 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 45996 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 45997 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 45998 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 45999 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 46000 + }, + { + "epoch": 2.79, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 46001 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 46002 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 46003 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.0845, + "step": 46004 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1877, + "step": 46005 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 46006 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.0916, + "step": 46007 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.0918, + "step": 46008 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 46009 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 46010 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 46011 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 46012 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.0822, + "step": 46013 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 46014 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 46015 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.0882, + "step": 46016 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 46017 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 46018 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 46019 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 46020 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 46021 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 46022 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.0916, + "step": 46023 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 46024 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 46025 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.0856, + "step": 46026 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 46027 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 46028 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 46029 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 46030 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 46031 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 46032 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 46033 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 46034 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 46035 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 46036 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 46037 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 46038 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 46039 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 46040 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1695, + "step": 46041 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.0943, + "step": 46042 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.0841, + "step": 46043 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 46044 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1741, + "step": 46045 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 46046 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.0876, + "step": 46047 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.0881, + "step": 46048 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 46049 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 46050 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 46051 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.0869, + "step": 46052 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 46053 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 46054 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 46055 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 46056 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 46057 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 46058 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 46059 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.165, + "step": 46060 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 46061 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.0839, + "step": 46062 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 46063 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 46064 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 46065 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 46066 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 46067 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 46068 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 46069 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 46070 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.0845, + "step": 46071 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 46072 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.0765, + "step": 46073 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 46074 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 46075 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 46076 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 46077 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 46078 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1614, + "step": 46079 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 46080 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 46081 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 46082 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 46083 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 46084 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 46085 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1507, + "step": 46086 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 46087 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 46088 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 46089 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 46090 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 46091 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 46092 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 46093 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.0903, + "step": 46094 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1548, + "step": 46095 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 46096 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 46097 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 46098 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 46099 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 46100 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1532, + "step": 46101 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 46102 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 46103 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 46104 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 46105 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 46106 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 46107 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 46108 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 46109 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 46110 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 46111 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 46112 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 46113 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 46114 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 46115 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 46116 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 46117 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 46118 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 46119 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 46120 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 46121 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 46122 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 46123 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 46124 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 46125 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 46126 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 46127 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 46128 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 46129 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 46130 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 46131 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 46132 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.0929, + "step": 46133 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1667, + "step": 46134 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.0859, + "step": 46135 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 46136 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 46137 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.0918, + "step": 46138 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 46139 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 46140 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 46141 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 46142 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 46143 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.085, + "step": 46144 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 46145 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 46146 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 46147 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 46148 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 46149 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 46150 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 46151 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 46152 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 46153 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 46154 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 46155 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 46156 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 46157 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 46158 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 46159 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 46160 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 46161 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 46162 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 46163 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 46164 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 46165 + }, + { + "epoch": 2.8, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 46166 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 46167 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 46168 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 46169 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 46170 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 46171 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 46172 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 46173 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.0899, + "step": 46174 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 46175 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.0897, + "step": 46176 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 46177 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 46178 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 46179 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1506, + "step": 46180 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 46181 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 46182 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 46183 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 46184 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.0864, + "step": 46185 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 46186 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 46187 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 46188 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 46189 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.089, + "step": 46190 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 46191 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 46192 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 46193 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 46194 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 46195 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.0906, + "step": 46196 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 46197 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 46198 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 46199 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 46200 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 46201 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 46202 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 46203 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 46204 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 46205 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 46206 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1503, + "step": 46207 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 46208 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 46209 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 46210 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 46211 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 46212 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 46213 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 46214 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 46215 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 46216 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 46217 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 46218 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 46219 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 46220 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 46221 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 46222 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 46223 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 46224 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 46225 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 46226 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 46227 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 46228 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 46229 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 46230 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 46231 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 46232 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 46233 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 46234 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 46235 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 46236 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 46237 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 46238 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 46239 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 46240 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 46241 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 46242 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 46243 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 46244 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.0915, + "step": 46245 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 46246 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 46247 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 46248 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 46249 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 46250 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 46251 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.0789, + "step": 46252 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 46253 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.09, + "step": 46254 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 46255 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.0932, + "step": 46256 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 46257 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 46258 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 46259 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 46260 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 46261 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 46262 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 46263 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 46264 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 46265 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.0757, + "step": 46266 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 46267 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.147, + "step": 46268 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.0878, + "step": 46269 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 46270 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 46271 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 46272 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 46273 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 46274 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 46275 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 46276 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 46277 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 46278 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 46279 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 46280 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 46281 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 46282 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 46283 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 46284 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 46285 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.0821, + "step": 46286 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 46287 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 46288 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 46289 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 46290 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 46291 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 46292 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 46293 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 46294 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.0807, + "step": 46295 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 46296 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 46297 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 46298 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.239, + "step": 46299 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 46300 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 46301 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 46302 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 46303 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 46304 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 46305 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 46306 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 46307 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 46308 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 46309 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 46310 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 46311 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 46312 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.0967, + "step": 46313 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 46314 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1429, + "step": 46315 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.173, + "step": 46316 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 46317 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 46318 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 46319 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 46320 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1555, + "step": 46321 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 46322 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 46323 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.0781, + "step": 46324 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 46325 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 46326 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 46327 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 46328 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 46329 + }, + { + "epoch": 2.81, + "learning_rate": 2e-05, + "loss": 0.0906, + "step": 46330 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 46331 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 46332 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 46333 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 46334 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 46335 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 46336 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 46337 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 46338 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.0868, + "step": 46339 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 46340 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 46341 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 46342 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 46343 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 46344 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 46345 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 46346 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 46347 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.0911, + "step": 46348 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 46349 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 46350 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.087, + "step": 46351 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 46352 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 46353 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 46354 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1696, + "step": 46355 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 46356 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 46357 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 46358 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 46359 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 46360 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 46361 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 46362 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 46363 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 46364 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 46365 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 46366 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 46367 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 46368 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 46369 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 46370 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 46371 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 46372 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 46373 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 46374 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 46375 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 46376 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 46377 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 46378 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 46379 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 46380 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 46381 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 46382 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.0872, + "step": 46383 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 46384 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 46385 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 46386 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 46387 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 46388 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 46389 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 46390 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 46391 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 46392 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 46393 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 46394 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 46395 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 46396 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 46397 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 46398 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 46399 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 46400 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.0762, + "step": 46401 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 46402 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 46403 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 46404 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 46405 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 46406 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 46407 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 46408 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1677, + "step": 46409 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 46410 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 46411 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 46412 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 46413 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 46414 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.0831, + "step": 46415 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 46416 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 46417 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.0936, + "step": 46418 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 46419 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 46420 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 46421 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 46422 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 46423 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 46424 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 46425 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 46426 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 46427 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 46428 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1643, + "step": 46429 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 46430 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 46431 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.0864, + "step": 46432 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 46433 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 46434 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 46435 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 46436 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 46437 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 46438 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 46439 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 46440 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 46441 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 46442 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 46443 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 46444 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 46445 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 46446 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 46447 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.0935, + "step": 46448 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 46449 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.163, + "step": 46450 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 46451 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 46452 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 46453 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 46454 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 46455 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 46456 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 46457 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 46458 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 46459 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 46460 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 46461 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 46462 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 46463 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 46464 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 46465 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 46466 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1387, + "step": 46467 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 46468 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 46469 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1511, + "step": 46470 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 46471 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 46472 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 46473 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 46474 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 46475 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 46476 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 46477 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 46478 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 46479 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 46480 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 46481 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.0935, + "step": 46482 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1579, + "step": 46483 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 46484 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 46485 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 46486 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 46487 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 46488 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 46489 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 46490 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 46491 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 46492 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1442, + "step": 46493 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 46494 + }, + { + "epoch": 2.82, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 46495 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 46496 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 46497 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 46498 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 46499 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 46500 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 46501 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 46502 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 46503 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 46504 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 46505 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 46506 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 46507 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 46508 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 46509 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 46510 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 46511 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 46512 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 46513 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 46514 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 46515 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 46516 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 46517 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 46518 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 46519 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 46520 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 46521 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 46522 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 46523 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 46524 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 46525 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.0847, + "step": 46526 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 46527 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 46528 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 46529 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1462, + "step": 46530 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1661, + "step": 46531 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 46532 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 46533 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 46534 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 46535 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 46536 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 46537 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 46538 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 46539 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.0941, + "step": 46540 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 46541 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 46542 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 46543 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 46544 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 46545 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 46546 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 46547 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 46548 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 46549 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 46550 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 46551 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 46552 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 46553 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.148, + "step": 46554 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 46555 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 46556 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1834, + "step": 46557 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 46558 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.0758, + "step": 46559 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 46560 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 46561 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 46562 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 46563 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 46564 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 46565 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 46566 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 46567 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 46568 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 46569 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 46570 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 46571 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 46572 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 46573 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 46574 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 46575 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 46576 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 46577 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 46578 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 46579 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 46580 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 46581 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 46582 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 46583 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 46584 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 46585 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 46586 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1517, + "step": 46587 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 46588 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 46589 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 46590 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 46591 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 46592 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 46593 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 46594 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 46595 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1456, + "step": 46596 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 46597 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 46598 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 46599 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1617, + "step": 46600 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 46601 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1511, + "step": 46602 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 46603 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 46604 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.0847, + "step": 46605 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.0889, + "step": 46606 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 46607 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 46608 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 46609 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 46610 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 46611 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 46612 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 46613 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 46614 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 46615 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 46616 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1447, + "step": 46617 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 46618 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 46619 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 46620 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 46621 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 46622 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 46623 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 46624 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 46625 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 46626 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.0917, + "step": 46627 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 46628 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.0866, + "step": 46629 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 46630 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 46631 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.0906, + "step": 46632 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1561, + "step": 46633 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 46634 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 46635 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 46636 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1663, + "step": 46637 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1523, + "step": 46638 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 46639 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 46640 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 46641 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 46642 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 46643 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 46644 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 46645 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 46646 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 46647 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 46648 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 46649 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 46650 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 46651 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 46652 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1518, + "step": 46653 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 46654 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 46655 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1852, + "step": 46656 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.0906, + "step": 46657 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 46658 + }, + { + "epoch": 2.83, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 46659 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 46660 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 46661 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 46662 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1429, + "step": 46663 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.0765, + "step": 46664 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 46665 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 46666 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 46667 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 46668 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 46669 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 46670 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 46671 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 46672 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 46673 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1641, + "step": 46674 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 46675 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 46676 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 46677 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 46678 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 46679 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 46680 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 46681 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 46682 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 46683 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 46684 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 46685 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 46686 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 46687 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 46688 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 46689 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.0917, + "step": 46690 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 46691 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 46692 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 46693 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 46694 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 46695 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 46696 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.165, + "step": 46697 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 46698 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 46699 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 46700 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 46701 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.0887, + "step": 46702 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 46703 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 46704 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 46705 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 46706 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 46707 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.0768, + "step": 46708 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 46709 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 46710 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 46711 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 46712 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 46713 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 46714 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 46715 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 46716 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 46717 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 46718 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 46719 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 46720 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 46721 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 46722 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 46723 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 46724 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 46725 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 46726 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 46727 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.0932, + "step": 46728 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 46729 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 46730 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 46731 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 46732 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 46733 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.0888, + "step": 46734 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 46735 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.0874, + "step": 46736 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1895, + "step": 46737 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 46738 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 46739 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 46740 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 46741 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 46742 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 46743 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 46744 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1586, + "step": 46745 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 46746 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 46747 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 46748 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.0943, + "step": 46749 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 46750 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 46751 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 46752 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 46753 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 46754 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 46755 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.154, + "step": 46756 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 46757 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 46758 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1598, + "step": 46759 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 46760 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 46761 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 46762 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 46763 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 46764 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 46765 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 46766 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 46767 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 46768 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 46769 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.0855, + "step": 46770 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 46771 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 46772 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1456, + "step": 46773 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 46774 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 46775 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 46776 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 46777 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.0768, + "step": 46778 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 46779 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 46780 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 46781 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 46782 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 46783 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 46784 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 46785 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 46786 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 46787 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 46788 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 46789 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 46790 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 46791 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 46792 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.0821, + "step": 46793 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 46794 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.0818, + "step": 46795 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 46796 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 46797 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 46798 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 46799 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 46800 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.0897, + "step": 46801 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 46802 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 46803 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 46804 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1531, + "step": 46805 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.0769, + "step": 46806 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 46807 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 46808 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 46809 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 46810 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 46811 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 46812 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1551, + "step": 46813 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 46814 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 46815 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 46816 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 46817 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.0918, + "step": 46818 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 46819 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 46820 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 46821 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 46822 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 46823 + }, + { + "epoch": 2.84, + "learning_rate": 2e-05, + "loss": 0.0851, + "step": 46824 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 46825 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 46826 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 46827 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 46828 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 46829 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.0867, + "step": 46830 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.0852, + "step": 46831 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 46832 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 46833 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 46834 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 46835 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 46836 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 46837 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 46838 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 46839 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 46840 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 46841 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 46842 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 46843 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 46844 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 46845 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 46846 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 46847 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 46848 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1585, + "step": 46849 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 46850 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 46851 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 46852 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 46853 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 46854 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 46855 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 46856 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 46857 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 46858 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 46859 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 46860 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 46861 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 46862 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 46863 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1553, + "step": 46864 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 46865 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.0873, + "step": 46866 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 46867 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 46868 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 46869 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 46870 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1497, + "step": 46871 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 46872 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 46873 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1488, + "step": 46874 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 46875 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 46876 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 46877 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 46878 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 46879 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.0936, + "step": 46880 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 46881 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 46882 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 46883 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 46884 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1449, + "step": 46885 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 46886 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 46887 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1497, + "step": 46888 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 46889 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 46890 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.0894, + "step": 46891 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 46892 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 46893 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.0887, + "step": 46894 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 46895 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 46896 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 46897 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 46898 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 46899 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 46900 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 46901 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1679, + "step": 46902 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 46903 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 46904 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 46905 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 46906 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 46907 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 46908 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 46909 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 46910 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 46911 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.0899, + "step": 46912 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 46913 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 46914 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1597, + "step": 46915 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.0821, + "step": 46916 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 46917 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.0892, + "step": 46918 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 46919 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 46920 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 46921 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 46922 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1798, + "step": 46923 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 46924 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 46925 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 46926 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 46927 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 46928 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 46929 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 46930 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 46931 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 46932 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 46933 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1478, + "step": 46934 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 46935 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 46936 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 46937 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.0903, + "step": 46938 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 46939 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 46940 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1616, + "step": 46941 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 46942 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 46943 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 46944 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 46945 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 46946 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 46947 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 46948 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 46949 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 46950 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 46951 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 46952 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 46953 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 46954 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.0877, + "step": 46955 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 46956 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1585, + "step": 46957 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 46958 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 46959 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 46960 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 46961 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 46962 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 46963 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1568, + "step": 46964 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 46965 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 46966 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 46967 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.089, + "step": 46968 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 46969 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 46970 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 46971 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 46972 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 46973 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 46974 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 46975 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 46976 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 46977 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 46978 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 46979 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.0835, + "step": 46980 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 46981 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 46982 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 46983 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 46984 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1596, + "step": 46985 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 46986 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 46987 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 46988 + }, + { + "epoch": 2.85, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 46989 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 46990 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 46991 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 46992 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 46993 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1697, + "step": 46994 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 46995 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.0894, + "step": 46996 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 46997 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 46998 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 46999 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 47000 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.0865, + "step": 47001 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 47002 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 47003 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 47004 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 47005 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 47006 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 47007 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 47008 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 47009 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 47010 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 47011 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.0873, + "step": 47012 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 47013 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1582, + "step": 47014 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 47015 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 47016 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.0871, + "step": 47017 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 47018 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 47019 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 47020 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 47021 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 47022 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 47023 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 47024 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 47025 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 47026 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 47027 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 47028 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 47029 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 47030 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 47031 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 47032 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 47033 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.0936, + "step": 47034 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 47035 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 47036 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 47037 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 47038 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 47039 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 47040 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1408, + "step": 47041 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 47042 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 47043 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 47044 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 47045 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 47046 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 47047 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 47048 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 47049 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.0865, + "step": 47050 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 47051 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 47052 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 47053 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 47054 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 47055 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.145, + "step": 47056 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 47057 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 47058 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 47059 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 47060 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 47061 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 47062 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.0911, + "step": 47063 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 47064 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 47065 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1653, + "step": 47066 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 47067 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 47068 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1612, + "step": 47069 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 47070 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 47071 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 47072 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1502, + "step": 47073 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 47074 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.0747, + "step": 47075 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 47076 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1559, + "step": 47077 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 47078 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 47079 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.0918, + "step": 47080 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 47081 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 47082 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 47083 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 47084 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 47085 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 47086 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 47087 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 47088 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 47089 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 47090 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 47091 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 47092 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 47093 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 47094 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 47095 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 47096 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 47097 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 47098 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 47099 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 47100 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 47101 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.0952, + "step": 47102 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 47103 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 47104 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 47105 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 47106 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 47107 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 47108 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 47109 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 47110 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 47111 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 47112 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 47113 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 47114 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 47115 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 47116 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 47117 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 47118 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 47119 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 47120 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 47121 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 47122 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 47123 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 47124 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 47125 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1452, + "step": 47126 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 47127 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 47128 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.0943, + "step": 47129 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 47130 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.0952, + "step": 47131 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 47132 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.0838, + "step": 47133 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 47134 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 47135 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 47136 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 47137 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 47138 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 47139 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 47140 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 47141 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 47142 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 47143 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 47144 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 47145 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 47146 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.0967, + "step": 47147 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 47148 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 47149 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 47150 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 47151 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 47152 + }, + { + "epoch": 2.86, + "learning_rate": 2e-05, + "loss": 0.0755, + "step": 47153 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 47154 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 47155 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 47156 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 47157 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 47158 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 47159 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 47160 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.0943, + "step": 47161 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 47162 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 47163 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1572, + "step": 47164 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 47165 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 47166 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 47167 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 47168 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 47169 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 47170 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 47171 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 47172 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1774, + "step": 47173 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 47174 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 47175 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 47176 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 47177 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 47178 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 47179 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 47180 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 47181 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 47182 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 47183 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 47184 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 47185 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 47186 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.0858, + "step": 47187 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 47188 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 47189 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.0902, + "step": 47190 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 47191 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 47192 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 47193 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 47194 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 47195 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 47196 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1627, + "step": 47197 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 47198 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 47199 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 47200 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 47201 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 47202 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 47203 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 47204 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 47205 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 47206 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 47207 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.0895, + "step": 47208 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 47209 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 47210 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 47211 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 47212 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 47213 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 47214 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1542, + "step": 47215 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 47216 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 47217 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 47218 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 47219 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.0881, + "step": 47220 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 47221 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 47222 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 47223 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 47224 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 47225 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 47226 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 47227 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 47228 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 47229 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 47230 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 47231 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 47232 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1562, + "step": 47233 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 47234 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 47235 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 47236 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 47237 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 47238 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 47239 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 47240 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.0793, + "step": 47241 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 47242 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 47243 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 47244 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 47245 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 47246 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 47247 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 47248 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 47249 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 47250 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 47251 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 47252 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.085, + "step": 47253 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 47254 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 47255 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 47256 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 47257 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 47258 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 47259 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.0941, + "step": 47260 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 47261 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 47262 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 47263 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 47264 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 47265 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 47266 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.0872, + "step": 47267 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 47268 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 47269 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1515, + "step": 47270 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 47271 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 47272 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 47273 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 47274 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 47275 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 47276 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 47277 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 47278 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1566, + "step": 47279 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 47280 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.148, + "step": 47281 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1764, + "step": 47282 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 47283 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 47284 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 47285 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 47286 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 47287 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1566, + "step": 47288 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 47289 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 47290 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 47291 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 47292 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 47293 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 47294 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 47295 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 47296 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 47297 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 47298 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 47299 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 47300 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 47301 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 47302 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 47303 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 47304 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 47305 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.0714, + "step": 47306 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 47307 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 47308 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 47309 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 47310 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 47311 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 47312 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 47313 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 47314 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1906, + "step": 47315 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 47316 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 47317 + }, + { + "epoch": 2.87, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 47318 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 47319 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 47320 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 47321 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 47322 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.0882, + "step": 47323 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 47324 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1502, + "step": 47325 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 47326 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 47327 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 47328 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1651, + "step": 47329 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 47330 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 47331 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 47332 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 47333 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 47334 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 47335 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 47336 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 47337 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 47338 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 47339 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 47340 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 47341 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 47342 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.16, + "step": 47343 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 47344 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 47345 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1647, + "step": 47346 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 47347 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.0846, + "step": 47348 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 47349 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 47350 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 47351 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 47352 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 47353 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 47354 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 47355 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.0914, + "step": 47356 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 47357 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 47358 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1442, + "step": 47359 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 47360 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 47361 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 47362 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 47363 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 47364 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 47365 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 47366 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 47367 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 47368 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 47369 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 47370 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 47371 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 47372 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 47373 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 47374 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 47375 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 47376 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 47377 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1583, + "step": 47378 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 47379 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 47380 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 47381 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 47382 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 47383 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.0819, + "step": 47384 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 47385 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 47386 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 47387 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 47388 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1673, + "step": 47389 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 47390 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 47391 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1812, + "step": 47392 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 47393 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 47394 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 47395 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 47396 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 47397 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 47398 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.0943, + "step": 47399 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 47400 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 47401 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 47402 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 47403 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 47404 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 47405 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 47406 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 47407 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 47408 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 47409 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 47410 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 47411 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 47412 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 47413 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 47414 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 47415 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 47416 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 47417 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.0864, + "step": 47418 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 47419 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 47420 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.0709, + "step": 47421 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.159, + "step": 47422 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 47423 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1429, + "step": 47424 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 47425 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 47426 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 47427 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 47428 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 47429 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 47430 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 47431 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 47432 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 47433 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 47434 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 47435 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 47436 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 47437 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 47438 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 47439 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 47440 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 47441 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 47442 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 47443 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 47444 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 47445 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 47446 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 47447 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 47448 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 47449 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 47450 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 47451 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 47452 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 47453 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1342, + "step": 47454 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 47455 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 47456 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 47457 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.0836, + "step": 47458 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 47459 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 47460 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 47461 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.0903, + "step": 47462 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 47463 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 47464 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 47465 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 47466 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 47467 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.07, + "step": 47468 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 47469 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 47470 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 47471 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 47472 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 47473 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 47474 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 47475 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 47476 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 47477 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 47478 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 47479 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 47480 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 47481 + }, + { + "epoch": 2.88, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 47482 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 47483 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 47484 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 47485 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 47486 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 47487 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 47488 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 47489 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 47490 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1689, + "step": 47491 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 47492 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 47493 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.168, + "step": 47494 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 47495 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 47496 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 47497 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 47498 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 47499 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 47500 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 47501 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.146, + "step": 47502 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 47503 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 47504 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 47505 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 47506 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 47507 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 47508 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 47509 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 47510 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 47511 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 47512 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 47513 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 47514 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 47515 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 47516 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 47517 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 47518 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 47519 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1467, + "step": 47520 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 47521 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 47522 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 47523 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 47524 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.0936, + "step": 47525 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 47526 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 47527 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 47528 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1423, + "step": 47529 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 47530 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 47531 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 47532 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 47533 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 47534 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 47535 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 47536 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 47537 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 47538 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 47539 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 47540 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 47541 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 47542 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 47543 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.0693, + "step": 47544 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 47545 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1476, + "step": 47546 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 47547 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.0845, + "step": 47548 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 47549 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.0867, + "step": 47550 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 47551 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1607, + "step": 47552 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 47553 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 47554 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 47555 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 47556 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 47557 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 47558 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 47559 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.0887, + "step": 47560 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 47561 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 47562 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 47563 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.0967, + "step": 47564 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 47565 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 47566 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 47567 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 47568 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 47569 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 47570 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 47571 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.0891, + "step": 47572 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 47573 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 47574 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 47575 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 47576 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.0887, + "step": 47577 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 47578 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 47579 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 47580 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 47581 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 47582 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.153, + "step": 47583 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 47584 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 47585 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.0874, + "step": 47586 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 47587 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 47588 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.0779, + "step": 47589 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 47590 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 47591 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1527, + "step": 47592 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 47593 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 47594 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 47595 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 47596 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 47597 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 47598 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 47599 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 47600 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 47601 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 47602 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 47603 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 47604 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 47605 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1532, + "step": 47606 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 47607 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 47608 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 47609 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 47610 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 47611 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 47612 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 47613 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 47614 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 47615 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 47616 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.0915, + "step": 47617 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 47618 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 47619 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 47620 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 47621 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 47622 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 47623 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 47624 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 47625 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 47626 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 47627 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 47628 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 47629 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 47630 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 47631 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 47632 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 47633 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 47634 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 47635 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 47636 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 47637 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.145, + "step": 47638 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 47639 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 47640 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 47641 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 47642 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 47643 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 47644 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 47645 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 47646 + }, + { + "epoch": 2.89, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 47647 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 47648 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 47649 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 47650 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 47651 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 47652 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1462, + "step": 47653 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 47654 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 47655 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 47656 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 47657 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 47658 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 47659 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 47660 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 47661 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.0831, + "step": 47662 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 47663 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 47664 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 47665 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 47666 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 47667 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 47668 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 47669 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1662, + "step": 47670 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 47671 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.0821, + "step": 47672 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 47673 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 47674 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 47675 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 47676 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.089, + "step": 47677 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 47678 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 47679 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 47680 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 47681 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 47682 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 47683 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 47684 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 47685 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 47686 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 47687 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 47688 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 47689 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 47690 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 47691 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 47692 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 47693 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 47694 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 47695 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 47696 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 47697 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 47698 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.0829, + "step": 47699 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1462, + "step": 47700 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 47701 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 47702 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 47703 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 47704 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 47705 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 47706 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 47707 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 47708 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 47709 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 47710 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 47711 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 47712 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1451, + "step": 47713 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 47714 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 47715 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 47716 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 47717 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 47718 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 47719 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1494, + "step": 47720 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 47721 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 47722 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 47723 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 47724 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 47725 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 47726 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1872, + "step": 47727 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 47728 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 47729 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 47730 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 47731 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.0915, + "step": 47732 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 47733 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 47734 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 47735 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 47736 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 47737 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 47738 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 47739 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 47740 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 47741 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 47742 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 47743 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 47744 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 47745 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 47746 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 47747 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 47748 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 47749 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 47750 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 47751 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 47752 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 47753 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 47754 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 47755 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 47756 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 47757 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 47758 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 47759 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.0918, + "step": 47760 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 47761 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 47762 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 47763 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1647, + "step": 47764 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 47765 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 47766 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 47767 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 47768 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 47769 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 47770 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 47771 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 47772 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 47773 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 47774 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 47775 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 47776 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 47777 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 47778 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.089, + "step": 47779 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 47780 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 47781 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 47782 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 47783 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 47784 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 47785 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 47786 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 47787 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 47788 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 47789 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 47790 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 47791 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 47792 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1619, + "step": 47793 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 47794 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 47795 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 47796 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 47797 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 47798 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 47799 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.168, + "step": 47800 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 47801 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 47802 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 47803 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 47804 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 47805 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.0873, + "step": 47806 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.0914, + "step": 47807 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 47808 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 47809 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.0929, + "step": 47810 + }, + { + "epoch": 2.9, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 47811 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 47812 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.164, + "step": 47813 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 47814 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.0868, + "step": 47815 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 47816 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.0798, + "step": 47817 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 47818 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 47819 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.0779, + "step": 47820 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 47821 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 47822 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 47823 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 47824 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 47825 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 47826 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 47827 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 47828 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 47829 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 47830 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 47831 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 47832 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 47833 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 47834 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 47835 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 47836 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 47837 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 47838 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1646, + "step": 47839 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 47840 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 47841 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 47842 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 47843 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 47844 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.0877, + "step": 47845 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.156, + "step": 47846 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 47847 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 47848 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 47849 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 47850 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 47851 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 47852 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 47853 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 47854 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 47855 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 47856 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 47857 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.0967, + "step": 47858 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 47859 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 47860 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 47861 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 47862 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 47863 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 47864 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 47865 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1655, + "step": 47866 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 47867 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 47868 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 47869 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 47870 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.0936, + "step": 47871 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 47872 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 47873 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 47874 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 47875 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 47876 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 47877 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 47878 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 47879 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 47880 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 47881 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 47882 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 47883 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 47884 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 47885 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 47886 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 47887 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 47888 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 47889 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.0771, + "step": 47890 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 47891 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 47892 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 47893 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 47894 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 47895 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 47896 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.0837, + "step": 47897 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 47898 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 47899 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 47900 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 47901 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1616, + "step": 47902 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 47903 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 47904 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1417, + "step": 47905 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 47906 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 47907 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 47908 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 47909 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 47910 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 47911 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 47912 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 47913 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 47914 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 47915 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 47916 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1465, + "step": 47917 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 47918 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 47919 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 47920 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 47921 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 47922 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 47923 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 47924 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 47925 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 47926 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 47927 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 47928 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 47929 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1595, + "step": 47930 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.0862, + "step": 47931 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 47932 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 47933 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.0833, + "step": 47934 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 47935 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 47936 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 47937 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 47938 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 47939 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 47940 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 47941 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 47942 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 47943 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 47944 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 47945 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 47946 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.142, + "step": 47947 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 47948 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 47949 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 47950 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 47951 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 47952 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 47953 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 47954 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 47955 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 47956 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 47957 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 47958 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 47959 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 47960 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.0824, + "step": 47961 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 47962 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 47963 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 47964 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.0917, + "step": 47965 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 47966 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 47967 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 47968 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 47969 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 47970 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.0679, + "step": 47971 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1797, + "step": 47972 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 47973 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1413, + "step": 47974 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 47975 + }, + { + "epoch": 2.91, + "learning_rate": 2e-05, + "loss": 0.0877, + "step": 47976 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1471, + "step": 47977 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 47978 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 47979 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 47980 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 47981 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 47982 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1517, + "step": 47983 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 47984 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 47985 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 47986 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 47987 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 47988 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.0894, + "step": 47989 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 47990 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 47991 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 47992 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 47993 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 47994 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 47995 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 47996 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1677, + "step": 47997 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 47998 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 47999 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 48000 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 48001 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 48002 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 48003 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 48004 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 48005 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 48006 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 48007 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 48008 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 48009 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 48010 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 48011 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 48012 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 48013 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 48014 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 48015 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 48016 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 48017 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1391, + "step": 48018 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 48019 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 48020 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 48021 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.0844, + "step": 48022 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1555, + "step": 48023 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 48024 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 48025 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 48026 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 48027 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 48028 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 48029 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 48030 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1729, + "step": 48031 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 48032 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.174, + "step": 48033 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 48034 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 48035 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 48036 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 48037 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 48038 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.15, + "step": 48039 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1409, + "step": 48040 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 48041 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 48042 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 48043 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 48044 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 48045 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 48046 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 48047 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 48048 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 48049 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 48050 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 48051 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 48052 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 48053 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 48054 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 48055 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 48056 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 48057 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 48058 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 48059 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.0763, + "step": 48060 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 48061 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 48062 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 48063 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 48064 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 48065 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 48066 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1513, + "step": 48067 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 48068 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 48069 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 48070 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 48071 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 48072 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1653, + "step": 48073 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 48074 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 48075 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 48076 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 48077 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 48078 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 48079 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 48080 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 48081 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 48082 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 48083 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 48084 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 48085 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 48086 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 48087 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 48088 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1567, + "step": 48089 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1613, + "step": 48090 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 48091 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 48092 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 48093 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 48094 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 48095 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 48096 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 48097 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 48098 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 48099 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.0906, + "step": 48100 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 48101 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.082, + "step": 48102 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 48103 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 48104 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 48105 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 48106 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 48107 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 48108 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 48109 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 48110 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 48111 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 48112 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 48113 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 48114 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 48115 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 48116 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 48117 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 48118 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 48119 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 48120 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 48121 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 48122 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 48123 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.0874, + "step": 48124 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 48125 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 48126 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 48127 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 48128 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 48129 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 48130 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 48131 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 48132 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 48133 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 48134 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 48135 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 48136 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 48137 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 48138 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.0935, + "step": 48139 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 48140 + }, + { + "epoch": 2.92, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 48141 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 48142 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 48143 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 48144 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 48145 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 48146 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 48147 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 48148 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 48149 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 48150 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 48151 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 48152 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 48153 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 48154 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 48155 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 48156 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 48157 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 48158 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 48159 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 48160 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 48161 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1594, + "step": 48162 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 48163 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 48164 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 48165 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 48166 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 48167 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 48168 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 48169 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1469, + "step": 48170 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1742, + "step": 48171 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 48172 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 48173 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 48174 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 48175 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 48176 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.078, + "step": 48177 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 48178 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 48179 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 48180 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 48181 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 48182 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1743, + "step": 48183 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 48184 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 48185 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 48186 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 48187 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 48188 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 48189 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 48190 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 48191 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 48192 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 48193 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 48194 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 48195 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 48196 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 48197 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 48198 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 48199 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 48200 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 48201 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 48202 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 48203 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 48204 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.0901, + "step": 48205 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 48206 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 48207 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 48208 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 48209 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 48210 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 48211 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 48212 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 48213 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 48214 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1617, + "step": 48215 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 48216 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 48217 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 48218 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 48219 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.0831, + "step": 48220 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 48221 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 48222 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 48223 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 48224 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1587, + "step": 48225 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 48226 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 48227 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 48228 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 48229 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 48230 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 48231 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 48232 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 48233 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 48234 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 48235 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 48236 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 48237 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 48238 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 48239 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 48240 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 48241 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 48242 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 48243 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 48244 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.162, + "step": 48245 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 48246 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 48247 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.174, + "step": 48248 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 48249 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 48250 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1896, + "step": 48251 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1601, + "step": 48252 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 48253 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 48254 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 48255 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 48256 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 48257 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1451, + "step": 48258 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1544, + "step": 48259 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 48260 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 48261 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 48262 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 48263 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 48264 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.091, + "step": 48265 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 48266 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 48267 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 48268 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 48269 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 48270 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 48271 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 48272 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 48273 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 48274 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 48275 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.0929, + "step": 48276 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 48277 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 48278 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 48279 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1429, + "step": 48280 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 48281 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 48282 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 48283 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 48284 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 48285 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 48286 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 48287 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 48288 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 48289 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 48290 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 48291 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 48292 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1349, + "step": 48293 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1545, + "step": 48294 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 48295 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 48296 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 48297 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 48298 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 48299 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 48300 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.0778, + "step": 48301 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 48302 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 48303 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.0846, + "step": 48304 + }, + { + "epoch": 2.93, + "learning_rate": 2e-05, + "loss": 0.1626, + "step": 48305 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 48306 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 48307 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 48308 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 48309 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1322, + "step": 48310 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 48311 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 48312 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 48313 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 48314 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 48315 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 48316 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1569, + "step": 48317 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 48318 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1535, + "step": 48319 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 48320 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 48321 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 48322 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 48323 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.0835, + "step": 48324 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.0854, + "step": 48325 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 48326 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 48327 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 48328 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 48329 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 48330 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 48331 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 48332 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 48333 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.0932, + "step": 48334 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 48335 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 48336 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 48337 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 48338 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 48339 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 48340 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 48341 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 48342 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 48343 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 48344 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.0967, + "step": 48345 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 48346 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.148, + "step": 48347 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 48348 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 48349 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 48350 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 48351 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 48352 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 48353 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 48354 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 48355 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 48356 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 48357 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 48358 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 48359 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 48360 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.0886, + "step": 48361 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 48362 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 48363 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 48364 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1618, + "step": 48365 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 48366 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1734, + "step": 48367 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 48368 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 48369 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1465, + "step": 48370 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.0892, + "step": 48371 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 48372 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 48373 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 48374 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 48375 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 48376 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 48377 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 48378 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 48379 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 48380 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 48381 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 48382 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 48383 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 48384 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 48385 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 48386 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 48387 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 48388 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 48389 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 48390 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 48391 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 48392 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 48393 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1487, + "step": 48394 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 48395 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 48396 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 48397 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 48398 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 48399 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.0772, + "step": 48400 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 48401 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 48402 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 48403 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 48404 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 48405 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.0858, + "step": 48406 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 48407 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 48408 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 48409 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 48410 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 48411 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 48412 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 48413 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 48414 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 48415 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 48416 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 48417 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 48418 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 48419 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 48420 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 48421 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 48422 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 48423 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 48424 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 48425 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.0796, + "step": 48426 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 48427 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 48428 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 48429 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 48430 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1458, + "step": 48431 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 48432 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 48433 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 48434 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 48435 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 48436 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 48437 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1517, + "step": 48438 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 48439 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 48440 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 48441 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 48442 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 48443 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1574, + "step": 48444 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 48445 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 48446 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 48447 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 48448 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 48449 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 48450 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 48451 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1624, + "step": 48452 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 48453 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 48454 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1633, + "step": 48455 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1877, + "step": 48456 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 48457 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 48458 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 48459 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 48460 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 48461 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1645, + "step": 48462 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 48463 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 48464 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 48465 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 48466 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.087, + "step": 48467 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 48468 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1641, + "step": 48469 + }, + { + "epoch": 2.94, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 48470 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.0911, + "step": 48471 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 48472 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 48473 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1507, + "step": 48474 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 48475 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 48476 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 48477 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.089, + "step": 48478 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 48479 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1541, + "step": 48480 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 48481 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 48482 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1486, + "step": 48483 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.152, + "step": 48484 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 48485 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 48486 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1542, + "step": 48487 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.0888, + "step": 48488 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.0892, + "step": 48489 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 48490 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 48491 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 48492 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 48493 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.0895, + "step": 48494 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 48495 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 48496 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 48497 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 48498 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 48499 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 48500 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 48501 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 48502 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1508, + "step": 48503 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 48504 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 48505 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 48506 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 48507 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1598, + "step": 48508 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 48509 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 48510 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 48511 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 48512 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 48513 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 48514 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 48515 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 48516 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 48517 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 48518 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 48519 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 48520 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 48521 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 48522 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 48523 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1681, + "step": 48524 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 48525 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 48526 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 48527 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 48528 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 48529 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 48530 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 48531 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 48532 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 48533 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 48534 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1443, + "step": 48535 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 48536 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 48537 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 48538 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 48539 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 48540 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 48541 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.144, + "step": 48542 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 48543 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 48544 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.0952, + "step": 48545 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 48546 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 48547 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 48548 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 48549 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 48550 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 48551 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 48552 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.0864, + "step": 48553 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 48554 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.151, + "step": 48555 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 48556 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 48557 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 48558 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 48559 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 48560 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 48561 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 48562 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 48563 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 48564 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 48565 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1613, + "step": 48566 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 48567 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.0916, + "step": 48568 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 48569 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 48570 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 48571 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 48572 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 48573 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.0869, + "step": 48574 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 48575 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 48576 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.145, + "step": 48577 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 48578 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 48579 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 48580 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 48581 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 48582 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 48583 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 48584 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.0935, + "step": 48585 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 48586 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 48587 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 48588 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 48589 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 48590 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 48591 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 48592 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 48593 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.145, + "step": 48594 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 48595 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 48596 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 48597 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 48598 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 48599 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 48600 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 48601 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 48602 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1714, + "step": 48603 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1643, + "step": 48604 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1514, + "step": 48605 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 48606 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 48607 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.0773, + "step": 48608 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 48609 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 48610 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 48611 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 48612 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 48613 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.089, + "step": 48614 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.0882, + "step": 48615 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 48616 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 48617 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 48618 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 48619 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1507, + "step": 48620 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 48621 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 48622 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 48623 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.09, + "step": 48624 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 48625 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 48626 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 48627 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 48628 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 48629 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 48630 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 48631 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 48632 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 48633 + }, + { + "epoch": 2.95, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 48634 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 48635 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 48636 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 48637 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 48638 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 48639 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 48640 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 48641 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 48642 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 48643 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 48644 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 48645 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 48646 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1447, + "step": 48647 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 48648 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.0825, + "step": 48649 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 48650 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 48651 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 48652 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 48653 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 48654 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1576, + "step": 48655 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 48656 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 48657 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.0825, + "step": 48658 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 48659 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 48660 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1757, + "step": 48661 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 48662 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 48663 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 48664 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 48665 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1595, + "step": 48666 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 48667 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 48668 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 48669 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 48670 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 48671 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 48672 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 48673 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 48674 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 48675 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.0853, + "step": 48676 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 48677 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 48678 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 48679 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 48680 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 48681 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 48682 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 48683 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 48684 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 48685 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.145, + "step": 48686 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 48687 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 48688 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 48689 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 48690 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 48691 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 48692 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 48693 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 48694 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 48695 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 48696 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 48697 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 48698 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1318, + "step": 48699 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 48700 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 48701 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 48702 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 48703 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1877, + "step": 48704 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 48705 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 48706 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 48707 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 48708 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 48709 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.0851, + "step": 48710 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.0815, + "step": 48711 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 48712 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 48713 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 48714 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 48715 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.0791, + "step": 48716 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 48717 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 48718 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 48719 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 48720 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 48721 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 48722 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 48723 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 48724 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 48725 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 48726 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1518, + "step": 48727 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 48728 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 48729 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1443, + "step": 48730 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 48731 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 48732 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 48733 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 48734 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 48735 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 48736 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 48737 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 48738 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 48739 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 48740 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 48741 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 48742 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.0825, + "step": 48743 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 48744 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 48745 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 48746 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 48747 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 48748 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.0935, + "step": 48749 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 48750 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 48751 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1646, + "step": 48752 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.0875, + "step": 48753 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 48754 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 48755 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 48756 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1636, + "step": 48757 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.152, + "step": 48758 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 48759 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 48760 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 48761 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 48762 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 48763 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 48764 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 48765 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 48766 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 48767 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.0813, + "step": 48768 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 48769 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.081, + "step": 48770 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 48771 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 48772 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 48773 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.0836, + "step": 48774 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 48775 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 48776 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 48777 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.0818, + "step": 48778 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.0874, + "step": 48779 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1456, + "step": 48780 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 48781 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 48782 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 48783 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.0835, + "step": 48784 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.0809, + "step": 48785 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1547, + "step": 48786 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 48787 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 48788 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 48789 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.16, + "step": 48790 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 48791 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 48792 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 48793 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 48794 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 48795 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 48796 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 48797 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 48798 + }, + { + "epoch": 2.96, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 48799 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 48800 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 48801 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1739, + "step": 48802 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 48803 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 48804 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 48805 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 48806 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 48807 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 48808 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 48809 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 48810 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 48811 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 48812 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 48813 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 48814 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 48815 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 48816 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 48817 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 48818 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1496, + "step": 48819 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 48820 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 48821 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 48822 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 48823 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 48824 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1771, + "step": 48825 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 48826 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 48827 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1556, + "step": 48828 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 48829 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1555, + "step": 48830 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 48831 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.0909, + "step": 48832 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 48833 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 48834 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 48835 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1555, + "step": 48836 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 48837 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 48838 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 48839 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 48840 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 48841 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1576, + "step": 48842 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 48843 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1558, + "step": 48844 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 48845 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 48846 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 48847 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1425, + "step": 48848 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 48849 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 48850 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1612, + "step": 48851 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 48852 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 48853 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 48854 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 48855 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 48856 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 48857 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 48858 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 48859 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 48860 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 48861 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 48862 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 48863 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 48864 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 48865 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 48866 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 48867 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 48868 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 48869 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 48870 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 48871 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 48872 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.0876, + "step": 48873 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 48874 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 48875 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1479, + "step": 48876 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.0836, + "step": 48877 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 48878 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 48879 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 48880 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 48881 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 48882 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 48883 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 48884 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 48885 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 48886 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 48887 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 48888 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 48889 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 48890 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 48891 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 48892 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 48893 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 48894 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 48895 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 48896 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 48897 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 48898 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 48899 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1429, + "step": 48900 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1517, + "step": 48901 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 48902 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 48903 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 48904 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 48905 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 48906 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 48907 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 48908 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 48909 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 48910 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.0935, + "step": 48911 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 48912 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 48913 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 48914 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1737, + "step": 48915 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 48916 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 48917 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 48918 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 48919 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 48920 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 48921 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 48922 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 48923 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 48924 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 48925 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 48926 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 48927 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 48928 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 48929 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 48930 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 48931 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 48932 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 48933 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.0909, + "step": 48934 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 48935 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 48936 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 48937 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 48938 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 48939 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 48940 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 48941 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 48942 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 48943 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 48944 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1531, + "step": 48945 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 48946 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 48947 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 48948 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 48949 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1427, + "step": 48950 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 48951 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 48952 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 48953 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 48954 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 48955 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 48956 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 48957 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 48958 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 48959 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 48960 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 48961 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 48962 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.0759, + "step": 48963 + }, + { + "epoch": 2.97, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 48964 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 48965 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 48966 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 48967 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 48968 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 48969 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 48970 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 48971 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 48972 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 48973 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 48974 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 48975 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 48976 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 48977 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 48978 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 48979 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1459, + "step": 48980 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 48981 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.0902, + "step": 48982 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 48983 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 48984 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 48985 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 48986 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 48987 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 48988 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 48989 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 48990 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 48991 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 48992 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.0891, + "step": 48993 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 48994 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 48995 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 48996 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1552, + "step": 48997 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 48998 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 48999 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 49000 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 49001 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 49002 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 49003 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 49004 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1615, + "step": 49005 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 49006 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 49007 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 49008 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 49009 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 49010 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.0906, + "step": 49011 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 49012 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 49013 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 49014 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 49015 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 49016 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 49017 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 49018 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 49019 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 49020 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 49021 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 49022 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 49023 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 49024 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 49025 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 49026 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 49027 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 49028 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 49029 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 49030 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 49031 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.0867, + "step": 49032 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 49033 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 49034 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 49035 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 49036 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 49037 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 49038 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 49039 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 49040 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 49041 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 49042 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 49043 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 49044 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 49045 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 49046 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 49047 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 49048 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 49049 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 49050 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1786, + "step": 49051 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 49052 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 49053 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 49054 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1634, + "step": 49055 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 49056 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 49057 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 49058 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 49059 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 49060 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 49061 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 49062 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 49063 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 49064 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 49065 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 49066 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 49067 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 49068 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.143, + "step": 49069 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 49070 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 49071 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 49072 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 49073 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 49074 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 49075 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 49076 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 49077 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 49078 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 49079 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 49080 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1597, + "step": 49081 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 49082 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 49083 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 49084 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 49085 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 49086 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 49087 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 49088 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.089, + "step": 49089 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 49090 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.075, + "step": 49091 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 49092 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1574, + "step": 49093 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 49094 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 49095 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 49096 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 49097 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 49098 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 49099 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 49100 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 49101 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 49102 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1492, + "step": 49103 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.0872, + "step": 49104 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 49105 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 49106 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 49107 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 49108 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 49109 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 49110 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 49111 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 49112 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 49113 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 49114 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 49115 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 49116 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 49117 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 49118 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 49119 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 49120 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 49121 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 49122 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1488, + "step": 49123 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.176, + "step": 49124 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 49125 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 49126 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 49127 + }, + { + "epoch": 2.98, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 49128 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 49129 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1502, + "step": 49130 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 49131 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 49132 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 49133 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.158, + "step": 49134 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 49135 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 49136 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 49137 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 49138 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 49139 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1643, + "step": 49140 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 49141 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 49142 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 49143 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.079, + "step": 49144 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 49145 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 49146 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 49147 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 49148 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.0843, + "step": 49149 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 49150 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 49151 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 49152 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 49153 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 49154 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.0941, + "step": 49155 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 49156 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 49157 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 49158 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 49159 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 49160 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 49161 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 49162 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 49163 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 49164 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 49165 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 49166 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 49167 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 49168 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 49169 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 49170 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 49171 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 49172 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 49173 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 49174 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 49175 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 49176 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 49177 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 49178 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 49179 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 49180 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 49181 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 49182 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 49183 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 49184 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 49185 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 49186 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 49187 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 49188 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 49189 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 49190 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 49191 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1566, + "step": 49192 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 49193 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 49194 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 49195 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 49196 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 49197 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 49198 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 49199 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 49200 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 49201 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 49202 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1478, + "step": 49203 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 49204 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 49205 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 49206 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 49207 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 49208 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 49209 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.0809, + "step": 49210 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 49211 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 49212 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 49213 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 49214 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1571, + "step": 49215 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 49216 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 49217 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 49218 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 49219 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 49220 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 49221 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.0892, + "step": 49222 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.148, + "step": 49223 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 49224 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1577, + "step": 49225 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 49226 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 49227 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 49228 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 49229 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 49230 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.0851, + "step": 49231 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 49232 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1484, + "step": 49233 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 49234 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 49235 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 49236 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.0899, + "step": 49237 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 49238 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.0894, + "step": 49239 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.0855, + "step": 49240 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 49241 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 49242 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 49243 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 49244 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1915, + "step": 49245 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 49246 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 49247 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.0866, + "step": 49248 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 49249 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 49250 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 49251 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 49252 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1513, + "step": 49253 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 49254 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 49255 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1673, + "step": 49256 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 49257 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 49258 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.0891, + "step": 49259 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 49260 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 49261 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1447, + "step": 49262 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 49263 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 49264 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1483, + "step": 49265 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1434, + "step": 49266 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 49267 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 49268 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1518, + "step": 49269 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 49270 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 49271 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 49272 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 49273 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 49274 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 49275 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 49276 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 49277 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 49278 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.0835, + "step": 49279 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1437, + "step": 49280 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 49281 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 49282 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 49283 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 49284 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 49285 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 49286 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 49287 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.0879, + "step": 49288 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.146, + "step": 49289 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 49290 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 49291 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 49292 + }, + { + "epoch": 2.99, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 49293 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 49294 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1469, + "step": 49295 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0818, + "step": 49296 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 49297 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 49298 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0762, + "step": 49299 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 49300 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1287, + "step": 49301 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 49302 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 49303 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 49304 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 49305 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 49306 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 49307 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 49308 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 49309 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0802, + "step": 49310 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 49311 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 49312 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 49313 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 49314 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 49315 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 49316 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 49317 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0811, + "step": 49318 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 49319 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0838, + "step": 49320 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 49321 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 49322 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 49323 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 49324 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 49325 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 49326 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 49327 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 49328 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 49329 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 49330 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0936, + "step": 49331 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 49332 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 49333 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 49334 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 49335 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1414, + "step": 49336 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 49337 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 49338 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 49339 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 49340 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 49341 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 49342 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 49343 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 49344 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 49345 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 49346 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 49347 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 49348 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 49349 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.17, + "step": 49350 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.2666, + "step": 49351 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 49352 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 49353 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.089, + "step": 49354 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 49355 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 49356 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 49357 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 49358 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 49359 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 49360 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 49361 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 49362 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.144, + "step": 49363 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 49364 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 49365 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 49366 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 49367 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 49368 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 49369 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 49370 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 49371 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 49372 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 49373 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 49374 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 49375 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 49376 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0872, + "step": 49377 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 49378 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0887, + "step": 49379 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 49380 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 49381 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.081, + "step": 49382 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 49383 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0738, + "step": 49384 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.077, + "step": 49385 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0911, + "step": 49386 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 49387 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0764, + "step": 49388 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 49389 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 49390 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 49391 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 49392 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.091, + "step": 49393 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 49394 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 49395 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0876, + "step": 49396 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.083, + "step": 49397 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0833, + "step": 49398 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 49399 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 49400 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.08, + "step": 49401 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 49402 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 49403 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0746, + "step": 49404 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0846, + "step": 49405 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 49406 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 49407 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 49408 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0883, + "step": 49409 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0806, + "step": 49410 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0707, + "step": 49411 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 49412 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 49413 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 49414 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 49415 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0844, + "step": 49416 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0747, + "step": 49417 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 49418 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0856, + "step": 49419 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 49420 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 49421 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0792, + "step": 49422 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0886, + "step": 49423 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0663, + "step": 49424 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 49425 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 49426 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0914, + "step": 49427 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 49428 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 49429 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0855, + "step": 49430 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 49431 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 49432 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 49433 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0902, + "step": 49434 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 49435 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 49436 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 49437 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 49438 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 49439 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 49440 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0852, + "step": 49441 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0818, + "step": 49442 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 49443 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 49444 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 49445 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.081, + "step": 49446 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0898, + "step": 49447 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 49448 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 49449 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 49450 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 49451 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 49452 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0761, + "step": 49453 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 49454 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 49455 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.0735, + "step": 49456 + }, + { + "epoch": 3.0, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 49457 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 49458 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.087, + "step": 49459 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0828, + "step": 49460 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 49461 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 49462 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 49463 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 49464 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 49465 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0739, + "step": 49466 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 49467 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0915, + "step": 49468 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0837, + "step": 49469 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 49470 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 49471 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 49472 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 49473 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 49474 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 49475 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 49476 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 49477 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0862, + "step": 49478 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 49479 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 49480 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 49481 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0851, + "step": 49482 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 49483 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0759, + "step": 49484 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 49485 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0878, + "step": 49486 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0792, + "step": 49487 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0829, + "step": 49488 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 49489 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0761, + "step": 49490 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0635, + "step": 49491 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 49492 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 49493 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 49494 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 49495 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0737, + "step": 49496 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 49497 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 49498 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 49499 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 49500 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0915, + "step": 49501 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 49502 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0853, + "step": 49503 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0697, + "step": 49504 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0801, + "step": 49505 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0943, + "step": 49506 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0723, + "step": 49507 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 49508 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0701, + "step": 49509 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 49510 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0911, + "step": 49511 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 49512 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 49513 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 49514 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0787, + "step": 49515 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0903, + "step": 49516 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 49517 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 49518 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 49519 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.077, + "step": 49520 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0809, + "step": 49521 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 49522 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.081, + "step": 49523 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 49524 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 49525 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0859, + "step": 49526 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 49527 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 49528 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 49529 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 49530 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 49531 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 49532 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 49533 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 49534 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 49535 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.085, + "step": 49536 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 49537 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0739, + "step": 49538 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 49539 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 49540 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 49541 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0733, + "step": 49542 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 49543 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 49544 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0665, + "step": 49545 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 49546 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.09, + "step": 49547 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 49548 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0825, + "step": 49549 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 49550 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 49551 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0767, + "step": 49552 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0866, + "step": 49553 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 49554 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 49555 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0894, + "step": 49556 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 49557 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0843, + "step": 49558 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 49559 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.084, + "step": 49560 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0799, + "step": 49561 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 49562 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.09, + "step": 49563 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 49564 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 49565 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 49566 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 49567 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 49568 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 49569 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 49570 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 49571 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 49572 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0778, + "step": 49573 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 49574 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0887, + "step": 49575 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 49576 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 49577 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0776, + "step": 49578 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 49579 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0675, + "step": 49580 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0892, + "step": 49581 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 49582 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 49583 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 49584 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 49585 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 49586 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 49587 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 49588 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 49589 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 49590 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0943, + "step": 49591 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.089, + "step": 49592 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 49593 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 49594 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 49595 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 49596 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 49597 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 49598 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 49599 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0902, + "step": 49600 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0804, + "step": 49601 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0917, + "step": 49602 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 49603 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 49604 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 49605 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 49606 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 49607 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0873, + "step": 49608 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 49609 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0839, + "step": 49610 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0822, + "step": 49611 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 49612 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0767, + "step": 49613 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 49614 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 49615 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.085, + "step": 49616 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 49617 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 49618 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 49619 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0805, + "step": 49620 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 49621 + }, + { + "epoch": 3.01, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 49622 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 49623 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0902, + "step": 49624 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 49625 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 49626 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 49627 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0723, + "step": 49628 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0734, + "step": 49629 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 49630 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 49631 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 49632 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0601, + "step": 49633 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.089, + "step": 49634 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 49635 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0876, + "step": 49636 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 49637 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 49638 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 49639 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0899, + "step": 49640 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0845, + "step": 49641 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 49642 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 49643 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 49644 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 49645 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0804, + "step": 49646 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 49647 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0764, + "step": 49648 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0832, + "step": 49649 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 49650 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 49651 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0895, + "step": 49652 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0793, + "step": 49653 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0882, + "step": 49654 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 49655 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 49656 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 49657 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0762, + "step": 49658 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 49659 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 49660 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 49661 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 49662 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0856, + "step": 49663 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 49664 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0767, + "step": 49665 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 49666 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 49667 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 49668 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.085, + "step": 49669 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 49670 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 49671 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 49672 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 49673 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 49674 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 49675 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 49676 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 49677 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0853, + "step": 49678 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0829, + "step": 49679 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 49680 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 49681 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0891, + "step": 49682 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 49683 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0822, + "step": 49684 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 49685 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 49686 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 49687 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0779, + "step": 49688 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 49689 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 49690 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0792, + "step": 49691 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 49692 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0786, + "step": 49693 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0664, + "step": 49694 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0671, + "step": 49695 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 49696 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0859, + "step": 49697 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0857, + "step": 49698 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 49699 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 49700 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 49701 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0847, + "step": 49702 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 49703 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 49704 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 49705 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 49706 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0779, + "step": 49707 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 49708 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 49709 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 49710 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0879, + "step": 49711 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.066, + "step": 49712 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 49713 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0899, + "step": 49714 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 49715 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 49716 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 49717 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 49718 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 49719 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 49720 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 49721 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0866, + "step": 49722 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0734, + "step": 49723 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 49724 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0867, + "step": 49725 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 49726 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 49727 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 49728 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 49729 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 49730 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0846, + "step": 49731 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 49732 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0801, + "step": 49733 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 49734 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 49735 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0895, + "step": 49736 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0799, + "step": 49737 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0917, + "step": 49738 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0836, + "step": 49739 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 49740 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 49741 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0872, + "step": 49742 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0754, + "step": 49743 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0756, + "step": 49744 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 49745 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1463, + "step": 49746 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0815, + "step": 49747 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 49748 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 49749 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 49750 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 49751 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 49752 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0838, + "step": 49753 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 49754 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0824, + "step": 49755 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 49756 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0887, + "step": 49757 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0895, + "step": 49758 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0915, + "step": 49759 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 49760 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 49761 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0743, + "step": 49762 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0819, + "step": 49763 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.078, + "step": 49764 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0841, + "step": 49765 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 49766 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0844, + "step": 49767 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0846, + "step": 49768 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0892, + "step": 49769 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 49770 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0863, + "step": 49771 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 49772 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 49773 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 49774 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 49775 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 49776 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0911, + "step": 49777 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0832, + "step": 49778 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 49779 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 49780 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 49781 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 49782 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 49783 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0929, + "step": 49784 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 49785 + }, + { + "epoch": 3.02, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 49786 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 49787 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 49788 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 49789 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0809, + "step": 49790 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 49791 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 49792 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0758, + "step": 49793 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0865, + "step": 49794 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 49795 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0841, + "step": 49796 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0699, + "step": 49797 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0845, + "step": 49798 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0879, + "step": 49799 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0734, + "step": 49800 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0805, + "step": 49801 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 49802 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0895, + "step": 49803 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0867, + "step": 49804 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 49805 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0888, + "step": 49806 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0809, + "step": 49807 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 49808 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 49809 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 49810 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 49811 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 49812 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 49813 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0803, + "step": 49814 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0804, + "step": 49815 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 49816 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0865, + "step": 49817 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0862, + "step": 49818 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 49819 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 49820 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 49821 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0857, + "step": 49822 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.091, + "step": 49823 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0901, + "step": 49824 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 49825 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 49826 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 49827 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 49828 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 49829 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 49830 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 49831 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0795, + "step": 49832 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 49833 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 49834 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0767, + "step": 49835 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 49836 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 49837 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0894, + "step": 49838 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 49839 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 49840 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0858, + "step": 49841 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.071, + "step": 49842 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 49843 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0863, + "step": 49844 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0935, + "step": 49845 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0906, + "step": 49846 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 49847 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 49848 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 49849 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.091, + "step": 49850 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 49851 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0858, + "step": 49852 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 49853 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 49854 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 49855 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 49856 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0881, + "step": 49857 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 49858 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0829, + "step": 49859 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0811, + "step": 49860 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 49861 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0879, + "step": 49862 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 49863 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 49864 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 49865 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 49866 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.08, + "step": 49867 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 49868 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 49869 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 49870 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 49871 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.079, + "step": 49872 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0747, + "step": 49873 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0876, + "step": 49874 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 49875 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0886, + "step": 49876 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0819, + "step": 49877 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 49878 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 49879 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 49880 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 49881 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0828, + "step": 49882 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 49883 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 49884 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 49885 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 49886 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0875, + "step": 49887 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0841, + "step": 49888 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 49889 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 49890 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 49891 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 49892 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 49893 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0803, + "step": 49894 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0832, + "step": 49895 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0867, + "step": 49896 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0835, + "step": 49897 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0723, + "step": 49898 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 49899 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0878, + "step": 49900 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0872, + "step": 49901 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 49902 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 49903 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 49904 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0785, + "step": 49905 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0871, + "step": 49906 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 49907 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0798, + "step": 49908 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 49909 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 49910 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.084, + "step": 49911 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0891, + "step": 49912 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0699, + "step": 49913 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 49914 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0749, + "step": 49915 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0859, + "step": 49916 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 49917 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 49918 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0847, + "step": 49919 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0941, + "step": 49920 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0883, + "step": 49921 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 49922 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 49923 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0687, + "step": 49924 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 49925 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 49926 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 49927 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 49928 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0892, + "step": 49929 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0768, + "step": 49930 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0857, + "step": 49931 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0898, + "step": 49932 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0709, + "step": 49933 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 49934 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 49935 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 49936 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 49937 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 49938 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0832, + "step": 49939 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.088, + "step": 49940 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 49941 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 49942 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 49943 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 49944 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 49945 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 49946 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.075, + "step": 49947 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 49948 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 49949 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 49950 + }, + { + "epoch": 3.03, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 49951 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0952, + "step": 49952 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0941, + "step": 49953 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0856, + "step": 49954 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 49955 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 49956 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 49957 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0918, + "step": 49958 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 49959 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 49960 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0834, + "step": 49961 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 49962 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 49963 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 49964 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 49965 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 49966 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0895, + "step": 49967 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 49968 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0857, + "step": 49969 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 49970 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 49971 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 49972 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 49973 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 49974 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 49975 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 49976 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 49977 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 49978 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 49979 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0716, + "step": 49980 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 49981 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0932, + "step": 49982 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0795, + "step": 49983 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0886, + "step": 49984 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0828, + "step": 49985 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 49986 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 49987 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0725, + "step": 49988 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 49989 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.076, + "step": 49990 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0771, + "step": 49991 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0889, + "step": 49992 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 49993 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0631, + "step": 49994 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0763, + "step": 49995 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0817, + "step": 49996 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.09, + "step": 49997 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.074, + "step": 49998 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 49999 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0917, + "step": 50000 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 50001 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0869, + "step": 50002 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0833, + "step": 50003 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 50004 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 50005 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 50006 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 50007 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 50008 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0664, + "step": 50009 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 50010 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0874, + "step": 50011 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 50012 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 50013 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 50014 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 50015 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0855, + "step": 50016 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 50017 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 50018 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 50019 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 50020 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0815, + "step": 50021 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0847, + "step": 50022 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 50023 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 50024 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0856, + "step": 50025 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 50026 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 50027 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0882, + "step": 50028 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 50029 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 50030 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0891, + "step": 50031 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 50032 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0772, + "step": 50033 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0818, + "step": 50034 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 50035 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 50036 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.082, + "step": 50037 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 50038 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 50039 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0827, + "step": 50040 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 50041 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0848, + "step": 50042 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0841, + "step": 50043 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0898, + "step": 50044 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0757, + "step": 50045 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 50046 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0759, + "step": 50047 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 50048 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 50049 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 50050 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 50051 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0666, + "step": 50052 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0929, + "step": 50053 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 50054 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0932, + "step": 50055 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 50056 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0673, + "step": 50057 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0914, + "step": 50058 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 50059 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0864, + "step": 50060 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0844, + "step": 50061 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0637, + "step": 50062 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.089, + "step": 50063 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 50064 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 50065 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0841, + "step": 50066 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 50067 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 50068 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 50069 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 50070 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0828, + "step": 50071 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0867, + "step": 50072 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 50073 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 50074 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 50075 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 50076 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0821, + "step": 50077 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 50078 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 50079 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0849, + "step": 50080 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.091, + "step": 50081 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.082, + "step": 50082 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 50083 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 50084 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0694, + "step": 50085 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0771, + "step": 50086 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0805, + "step": 50087 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0771, + "step": 50088 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 50089 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0889, + "step": 50090 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 50091 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0866, + "step": 50092 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 50093 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 50094 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 50095 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0844, + "step": 50096 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 50097 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0844, + "step": 50098 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 50099 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 50100 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 50101 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 50102 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0779, + "step": 50103 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 50104 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 50105 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 50106 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 50107 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0853, + "step": 50108 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0684, + "step": 50109 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 50110 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 50111 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 50112 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 50113 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0848, + "step": 50114 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0806, + "step": 50115 + }, + { + "epoch": 3.04, + "learning_rate": 2e-05, + "loss": 0.0794, + "step": 50116 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 50117 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 50118 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 50119 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 50120 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0882, + "step": 50121 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 50122 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 50123 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 50124 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 50125 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 50126 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 50127 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 50128 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 50129 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0717, + "step": 50130 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 50131 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0936, + "step": 50132 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0894, + "step": 50133 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 50134 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0887, + "step": 50135 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0801, + "step": 50136 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 50137 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0878, + "step": 50138 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 50139 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 50140 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 50141 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 50142 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 50143 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 50144 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.074, + "step": 50145 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 50146 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 50147 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0887, + "step": 50148 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 50149 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 50150 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0868, + "step": 50151 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0797, + "step": 50152 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 50153 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 50154 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0895, + "step": 50155 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 50156 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0791, + "step": 50157 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 50158 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 50159 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0812, + "step": 50160 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 50161 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 50162 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 50163 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0836, + "step": 50164 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 50165 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 50166 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 50167 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 50168 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0722, + "step": 50169 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 50170 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0897, + "step": 50171 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 50172 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0822, + "step": 50173 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0813, + "step": 50174 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0743, + "step": 50175 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 50176 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 50177 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0657, + "step": 50178 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.089, + "step": 50179 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0784, + "step": 50180 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 50181 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0915, + "step": 50182 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0871, + "step": 50183 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 50184 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0845, + "step": 50185 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 50186 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 50187 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0943, + "step": 50188 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 50189 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0834, + "step": 50190 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 50191 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 50192 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 50193 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 50194 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0821, + "step": 50195 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 50196 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 50197 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0841, + "step": 50198 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1466, + "step": 50199 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 50200 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 50201 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0803, + "step": 50202 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 50203 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0911, + "step": 50204 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0859, + "step": 50205 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 50206 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 50207 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0836, + "step": 50208 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 50209 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 50210 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0849, + "step": 50211 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0721, + "step": 50212 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 50213 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0892, + "step": 50214 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 50215 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0906, + "step": 50216 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0935, + "step": 50217 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 50218 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 50219 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0807, + "step": 50220 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.084, + "step": 50221 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.088, + "step": 50222 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 50223 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 50224 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 50225 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 50226 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0836, + "step": 50227 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 50228 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 50229 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 50230 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0729, + "step": 50231 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 50232 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 50233 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0766, + "step": 50234 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 50235 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 50236 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 50237 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 50238 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 50239 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 50240 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0827, + "step": 50241 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 50242 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0837, + "step": 50243 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0773, + "step": 50244 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0638, + "step": 50245 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0805, + "step": 50246 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 50247 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 50248 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 50249 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 50250 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 50251 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 50252 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 50253 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 50254 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 50255 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 50256 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 50257 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.074, + "step": 50258 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 50259 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 50260 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 50261 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 50262 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 50263 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 50264 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0901, + "step": 50265 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 50266 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 50267 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0881, + "step": 50268 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 50269 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 50270 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 50271 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 50272 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 50273 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0771, + "step": 50274 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 50275 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0887, + "step": 50276 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.09, + "step": 50277 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 50278 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 50279 + }, + { + "epoch": 3.05, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 50280 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0876, + "step": 50281 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 50282 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 50283 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 50284 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0836, + "step": 50285 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 50286 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0936, + "step": 50287 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 50288 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0789, + "step": 50289 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 50290 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 50291 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.085, + "step": 50292 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 50293 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 50294 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 50295 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0918, + "step": 50296 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 50297 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 50298 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0894, + "step": 50299 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 50300 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 50301 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 50302 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 50303 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 50304 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0898, + "step": 50305 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0809, + "step": 50306 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 50307 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 50308 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 50309 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0806, + "step": 50310 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0733, + "step": 50311 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0848, + "step": 50312 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 50313 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 50314 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0799, + "step": 50315 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 50316 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.088, + "step": 50317 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0811, + "step": 50318 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0772, + "step": 50319 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 50320 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.09, + "step": 50321 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 50322 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0887, + "step": 50323 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.088, + "step": 50324 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 50325 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 50326 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 50327 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 50328 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 50329 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.073, + "step": 50330 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 50331 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 50332 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0701, + "step": 50333 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 50334 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 50335 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 50336 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 50337 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0902, + "step": 50338 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 50339 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 50340 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.079, + "step": 50341 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 50342 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0711, + "step": 50343 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0809, + "step": 50344 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 50345 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 50346 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0743, + "step": 50347 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 50348 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 50349 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0873, + "step": 50350 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0849, + "step": 50351 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 50352 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 50353 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0845, + "step": 50354 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 50355 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 50356 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0814, + "step": 50357 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.091, + "step": 50358 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 50359 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0807, + "step": 50360 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 50361 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 50362 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 50363 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0862, + "step": 50364 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0895, + "step": 50365 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 50366 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 50367 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 50368 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 50369 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0891, + "step": 50370 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 50371 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 50372 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 50373 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0857, + "step": 50374 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0778, + "step": 50375 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 50376 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 50377 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 50378 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 50379 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0826, + "step": 50380 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0815, + "step": 50381 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0839, + "step": 50382 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 50383 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 50384 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 50385 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 50386 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0855, + "step": 50387 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.091, + "step": 50388 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 50389 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0797, + "step": 50390 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 50391 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 50392 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 50393 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 50394 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0813, + "step": 50395 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 50396 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 50397 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0727, + "step": 50398 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 50399 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0917, + "step": 50400 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 50401 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0874, + "step": 50402 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0876, + "step": 50403 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0737, + "step": 50404 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 50405 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 50406 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0929, + "step": 50407 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0862, + "step": 50408 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 50409 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0883, + "step": 50410 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0864, + "step": 50411 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 50412 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0903, + "step": 50413 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 50414 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0675, + "step": 50415 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0848, + "step": 50416 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0836, + "step": 50417 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 50418 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 50419 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.09, + "step": 50420 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0715, + "step": 50421 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 50422 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.081, + "step": 50423 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 50424 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 50425 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 50426 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 50427 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 50428 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 50429 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0849, + "step": 50430 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0817, + "step": 50431 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 50432 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 50433 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 50434 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.081, + "step": 50435 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0914, + "step": 50436 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 50437 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 50438 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 50439 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0866, + "step": 50440 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 50441 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 50442 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 50443 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 50444 + }, + { + "epoch": 3.06, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 50445 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 50446 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 50447 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0906, + "step": 50448 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 50449 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0724, + "step": 50450 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 50451 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 50452 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0742, + "step": 50453 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 50454 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 50455 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 50456 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0714, + "step": 50457 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 50458 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 50459 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 50460 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0806, + "step": 50461 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 50462 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 50463 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 50464 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 50465 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0897, + "step": 50466 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 50467 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 50468 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 50469 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.09, + "step": 50470 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 50471 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 50472 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 50473 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 50474 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 50475 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0816, + "step": 50476 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 50477 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0879, + "step": 50478 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 50479 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0726, + "step": 50480 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0828, + "step": 50481 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 50482 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.073, + "step": 50483 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0648, + "step": 50484 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 50485 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 50486 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 50487 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 50488 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 50489 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 50490 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 50491 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 50492 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 50493 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 50494 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0879, + "step": 50495 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 50496 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 50497 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0758, + "step": 50498 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 50499 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 50500 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 50501 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0764, + "step": 50502 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0831, + "step": 50503 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 50504 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 50505 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0952, + "step": 50506 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 50507 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 50508 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0889, + "step": 50509 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0772, + "step": 50510 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0935, + "step": 50511 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 50512 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 50513 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 50514 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 50515 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0825, + "step": 50516 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0689, + "step": 50517 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 50518 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0816, + "step": 50519 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0881, + "step": 50520 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0807, + "step": 50521 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 50522 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0833, + "step": 50523 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0873, + "step": 50524 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 50525 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 50526 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0847, + "step": 50527 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 50528 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 50529 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0808, + "step": 50530 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 50531 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 50532 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 50533 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0866, + "step": 50534 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 50535 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 50536 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0776, + "step": 50537 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 50538 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 50539 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 50540 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 50541 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 50542 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0822, + "step": 50543 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 50544 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0935, + "step": 50545 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 50546 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0766, + "step": 50547 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0932, + "step": 50548 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0902, + "step": 50549 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 50550 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 50551 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 50552 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 50553 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 50554 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 50555 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0836, + "step": 50556 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 50557 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 50558 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 50559 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 50560 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 50561 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 50562 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 50563 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 50564 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 50565 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 50566 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 50567 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 50568 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0717, + "step": 50569 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0786, + "step": 50570 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0903, + "step": 50571 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0863, + "step": 50572 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0793, + "step": 50573 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 50574 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 50575 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 50576 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 50577 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 50578 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0765, + "step": 50579 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0732, + "step": 50580 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0841, + "step": 50581 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0827, + "step": 50582 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 50583 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0865, + "step": 50584 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0828, + "step": 50585 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 50586 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 50587 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 50588 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0696, + "step": 50589 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 50590 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 50591 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 50592 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 50593 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0834, + "step": 50594 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0883, + "step": 50595 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0802, + "step": 50596 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0918, + "step": 50597 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 50598 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 50599 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 50600 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0813, + "step": 50601 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 50602 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 50603 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0821, + "step": 50604 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 50605 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 50606 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 50607 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 50608 + }, + { + "epoch": 3.07, + "learning_rate": 2e-05, + "loss": 0.0838, + "step": 50609 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0858, + "step": 50610 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.077, + "step": 50611 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.078, + "step": 50612 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 50613 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 50614 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 50615 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0855, + "step": 50616 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 50617 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 50618 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 50619 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 50620 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0868, + "step": 50621 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.087, + "step": 50622 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 50623 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 50624 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0783, + "step": 50625 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0887, + "step": 50626 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.091, + "step": 50627 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 50628 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 50629 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 50630 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 50631 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.088, + "step": 50632 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 50633 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 50634 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0864, + "step": 50635 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 50636 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 50637 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 50638 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0887, + "step": 50639 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 50640 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0881, + "step": 50641 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 50642 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 50643 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 50644 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 50645 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0881, + "step": 50646 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0888, + "step": 50647 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 50648 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 50649 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 50650 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 50651 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 50652 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 50653 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 50654 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 50655 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 50656 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 50657 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0834, + "step": 50658 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0888, + "step": 50659 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 50660 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 50661 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0892, + "step": 50662 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 50663 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0848, + "step": 50664 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 50665 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 50666 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 50667 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 50668 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 50669 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 50670 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0854, + "step": 50671 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 50672 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 50673 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0858, + "step": 50674 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 50675 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 50676 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 50677 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0803, + "step": 50678 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 50679 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 50680 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 50681 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 50682 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0801, + "step": 50683 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.088, + "step": 50684 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 50685 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 50686 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0887, + "step": 50687 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 50688 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 50689 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0768, + "step": 50690 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 50691 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 50692 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 50693 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 50694 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 50695 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0872, + "step": 50696 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 50697 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0653, + "step": 50698 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0773, + "step": 50699 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 50700 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0883, + "step": 50701 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 50702 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 50703 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 50704 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 50705 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 50706 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 50707 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 50708 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 50709 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0864, + "step": 50710 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 50711 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0898, + "step": 50712 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0796, + "step": 50713 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 50714 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0858, + "step": 50715 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 50716 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0832, + "step": 50717 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 50718 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0862, + "step": 50719 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0815, + "step": 50720 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 50721 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 50722 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0783, + "step": 50723 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 50724 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.073, + "step": 50725 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 50726 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 50727 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0918, + "step": 50728 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0845, + "step": 50729 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0889, + "step": 50730 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 50731 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 50732 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0787, + "step": 50733 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 50734 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0879, + "step": 50735 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 50736 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.082, + "step": 50737 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 50738 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0898, + "step": 50739 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 50740 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0788, + "step": 50741 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0831, + "step": 50742 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.091, + "step": 50743 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 50744 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 50745 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0819, + "step": 50746 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 50747 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 50748 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 50749 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 50750 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 50751 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 50752 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0849, + "step": 50753 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 50754 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0831, + "step": 50755 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 50756 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 50757 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0866, + "step": 50758 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 50759 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 50760 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.075, + "step": 50761 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 50762 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0749, + "step": 50763 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0752, + "step": 50764 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 50765 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0851, + "step": 50766 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 50767 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.0809, + "step": 50768 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.065, + "step": 50769 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 50770 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 50771 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.083, + "step": 50772 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 50773 + }, + { + "epoch": 3.08, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 50774 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 50775 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0614, + "step": 50776 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 50777 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 50778 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 50779 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 50780 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0879, + "step": 50781 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 50782 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0834, + "step": 50783 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.085, + "step": 50784 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 50785 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0874, + "step": 50786 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 50787 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 50788 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0909, + "step": 50789 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1464, + "step": 50790 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 50791 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0873, + "step": 50792 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 50793 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0811, + "step": 50794 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0807, + "step": 50795 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0782, + "step": 50796 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.084, + "step": 50797 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0838, + "step": 50798 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 50799 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0967, + "step": 50800 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 50801 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0887, + "step": 50802 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 50803 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 50804 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 50805 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 50806 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 50807 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 50808 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0761, + "step": 50809 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0835, + "step": 50810 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 50811 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0787, + "step": 50812 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0799, + "step": 50813 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.081, + "step": 50814 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 50815 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 50816 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 50817 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 50818 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 50819 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 50820 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 50821 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 50822 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0915, + "step": 50823 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0866, + "step": 50824 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0828, + "step": 50825 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 50826 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0911, + "step": 50827 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 50828 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0862, + "step": 50829 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0887, + "step": 50830 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 50831 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0857, + "step": 50832 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 50833 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 50834 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0888, + "step": 50835 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 50836 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 50837 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0873, + "step": 50838 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 50839 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 50840 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0868, + "step": 50841 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0747, + "step": 50842 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 50843 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 50844 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0887, + "step": 50845 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 50846 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 50847 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 50848 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 50849 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 50850 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 50851 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 50852 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 50853 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0879, + "step": 50854 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 50855 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 50856 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 50857 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 50858 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0903, + "step": 50859 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 50860 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 50861 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 50862 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 50863 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 50864 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0883, + "step": 50865 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 50866 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0821, + "step": 50867 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0749, + "step": 50868 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 50869 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0735, + "step": 50870 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0755, + "step": 50871 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.078, + "step": 50872 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 50873 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 50874 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 50875 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0792, + "step": 50876 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 50877 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0783, + "step": 50878 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 50879 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 50880 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0803, + "step": 50881 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 50882 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0782, + "step": 50883 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 50884 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 50885 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0901, + "step": 50886 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 50887 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0808, + "step": 50888 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.081, + "step": 50889 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0865, + "step": 50890 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.073, + "step": 50891 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 50892 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 50893 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0859, + "step": 50894 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0756, + "step": 50895 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 50896 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0695, + "step": 50897 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0725, + "step": 50898 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 50899 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0816, + "step": 50900 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0631, + "step": 50901 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 50902 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0852, + "step": 50903 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 50904 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0812, + "step": 50905 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 50906 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0847, + "step": 50907 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 50908 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0872, + "step": 50909 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 50910 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 50911 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 50912 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 50913 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 50914 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 50915 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 50916 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 50917 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 50918 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0892, + "step": 50919 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 50920 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 50921 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 50922 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0889, + "step": 50923 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0774, + "step": 50924 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0879, + "step": 50925 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 50926 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 50927 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 50928 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 50929 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 50930 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 50931 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 50932 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0749, + "step": 50933 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 50934 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.077, + "step": 50935 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 50936 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 50937 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0911, + "step": 50938 + }, + { + "epoch": 3.09, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 50939 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 50940 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 50941 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 50942 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0967, + "step": 50943 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 50944 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 50945 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 50946 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 50947 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 50948 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 50949 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0846, + "step": 50950 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0836, + "step": 50951 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0859, + "step": 50952 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0842, + "step": 50953 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0875, + "step": 50954 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0849, + "step": 50955 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 50956 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 50957 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 50958 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0769, + "step": 50959 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 50960 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0895, + "step": 50961 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0828, + "step": 50962 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 50963 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 50964 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 50965 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 50966 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0833, + "step": 50967 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 50968 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0929, + "step": 50969 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0889, + "step": 50970 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0914, + "step": 50971 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0889, + "step": 50972 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0796, + "step": 50973 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 50974 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 50975 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 50976 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 50977 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0897, + "step": 50978 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 50979 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0751, + "step": 50980 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0886, + "step": 50981 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 50982 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 50983 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 50984 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 50985 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 50986 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 50987 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 50988 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 50989 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 50990 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0764, + "step": 50991 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 50992 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 50993 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 50994 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 50995 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 50996 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 50997 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0895, + "step": 50998 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 50999 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 51000 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 51001 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 51002 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0696, + "step": 51003 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 51004 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 51005 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 51006 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 51007 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 51008 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 51009 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0822, + "step": 51010 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 51011 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 51012 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 51013 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 51014 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 51015 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 51016 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0679, + "step": 51017 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0738, + "step": 51018 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0877, + "step": 51019 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 51020 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 51021 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 51022 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0952, + "step": 51023 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0799, + "step": 51024 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0879, + "step": 51025 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0864, + "step": 51026 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 51027 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 51028 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0868, + "step": 51029 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 51030 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 51031 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0872, + "step": 51032 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 51033 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 51034 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 51035 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.084, + "step": 51036 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 51037 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 51038 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 51039 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 51040 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 51041 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 51042 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 51043 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 51044 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 51045 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0763, + "step": 51046 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 51047 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 51048 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0821, + "step": 51049 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 51050 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 51051 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0775, + "step": 51052 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 51053 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 51054 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0809, + "step": 51055 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0744, + "step": 51056 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0894, + "step": 51057 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0839, + "step": 51058 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 51059 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0752, + "step": 51060 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 51061 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 51062 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 51063 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 51064 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 51065 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0889, + "step": 51066 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 51067 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 51068 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 51069 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 51070 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0852, + "step": 51071 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 51072 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 51073 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0932, + "step": 51074 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 51075 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0823, + "step": 51076 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.065, + "step": 51077 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0874, + "step": 51078 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 51079 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0932, + "step": 51080 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 51081 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 51082 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 51083 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 51084 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0834, + "step": 51085 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 51086 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0836, + "step": 51087 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0916, + "step": 51088 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0765, + "step": 51089 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0917, + "step": 51090 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 51091 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 51092 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 51093 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0712, + "step": 51094 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 51095 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 51096 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 51097 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 51098 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 51099 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 51100 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.0866, + "step": 51101 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 51102 + }, + { + "epoch": 3.1, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 51103 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0788, + "step": 51104 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 51105 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0906, + "step": 51106 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0837, + "step": 51107 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0883, + "step": 51108 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0829, + "step": 51109 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 51110 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 51111 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 51112 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 51113 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 51114 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0935, + "step": 51115 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 51116 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0878, + "step": 51117 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 51118 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 51119 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 51120 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 51121 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 51122 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 51123 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 51124 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.084, + "step": 51125 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 51126 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 51127 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 51128 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 51129 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 51130 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 51131 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0895, + "step": 51132 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 51133 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 51134 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0823, + "step": 51135 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0792, + "step": 51136 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 51137 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 51138 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 51139 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 51140 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0853, + "step": 51141 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0967, + "step": 51142 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 51143 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0936, + "step": 51144 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 51145 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 51146 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 51147 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 51148 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0808, + "step": 51149 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 51150 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0886, + "step": 51151 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0715, + "step": 51152 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 51153 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 51154 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 51155 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 51156 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0878, + "step": 51157 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 51158 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 51159 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0797, + "step": 51160 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0943, + "step": 51161 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0889, + "step": 51162 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 51163 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.082, + "step": 51164 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 51165 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 51166 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 51167 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 51168 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0723, + "step": 51169 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 51170 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0752, + "step": 51171 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 51172 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 51173 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 51174 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 51175 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 51176 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 51177 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0916, + "step": 51178 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0793, + "step": 51179 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 51180 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0826, + "step": 51181 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 51182 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 51183 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 51184 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 51185 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 51186 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0835, + "step": 51187 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.083, + "step": 51188 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 51189 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 51190 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 51191 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0708, + "step": 51192 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0881, + "step": 51193 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 51194 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 51195 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 51196 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0929, + "step": 51197 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 51198 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0898, + "step": 51199 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 51200 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 51201 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0918, + "step": 51202 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 51203 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0713, + "step": 51204 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 51205 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 51206 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0899, + "step": 51207 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 51208 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 51209 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0832, + "step": 51210 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 51211 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0932, + "step": 51212 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 51213 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 51214 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 51215 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 51216 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0851, + "step": 51217 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0887, + "step": 51218 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 51219 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 51220 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 51221 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 51222 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 51223 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 51224 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0917, + "step": 51225 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0878, + "step": 51226 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 51227 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0752, + "step": 51228 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 51229 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0902, + "step": 51230 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 51231 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0697, + "step": 51232 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0941, + "step": 51233 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 51234 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0708, + "step": 51235 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0809, + "step": 51236 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 51237 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 51238 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 51239 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 51240 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 51241 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 51242 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 51243 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 51244 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.075, + "step": 51245 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 51246 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 51247 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 51248 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0765, + "step": 51249 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0856, + "step": 51250 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 51251 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0929, + "step": 51252 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0575, + "step": 51253 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0899, + "step": 51254 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 51255 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 51256 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 51257 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0823, + "step": 51258 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 51259 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1525, + "step": 51260 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 51261 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 51262 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0841, + "step": 51263 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0804, + "step": 51264 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0838, + "step": 51265 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 51266 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.09, + "step": 51267 + }, + { + "epoch": 3.11, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 51268 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0862, + "step": 51269 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 51270 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 51271 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 51272 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0708, + "step": 51273 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0874, + "step": 51274 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 51275 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 51276 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 51277 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0894, + "step": 51278 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0877, + "step": 51279 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 51280 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 51281 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 51282 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 51283 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0759, + "step": 51284 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 51285 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 51286 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 51287 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0876, + "step": 51288 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0867, + "step": 51289 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0902, + "step": 51290 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 51291 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 51292 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 51293 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0776, + "step": 51294 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 51295 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 51296 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 51297 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0871, + "step": 51298 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 51299 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0827, + "step": 51300 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 51301 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 51302 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0726, + "step": 51303 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 51304 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 51305 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 51306 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0816, + "step": 51307 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0793, + "step": 51308 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 51309 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 51310 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.087, + "step": 51311 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 51312 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 51313 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0839, + "step": 51314 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0851, + "step": 51315 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 51316 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0829, + "step": 51317 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 51318 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0742, + "step": 51319 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 51320 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 51321 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0784, + "step": 51322 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 51323 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0859, + "step": 51324 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 51325 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0898, + "step": 51326 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0821, + "step": 51327 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0914, + "step": 51328 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0807, + "step": 51329 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 51330 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 51331 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 51332 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0871, + "step": 51333 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 51334 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0707, + "step": 51335 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 51336 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 51337 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 51338 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 51339 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 51340 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 51341 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0941, + "step": 51342 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 51343 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 51344 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0932, + "step": 51345 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1363, + "step": 51346 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 51347 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 51348 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 51349 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 51350 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 51351 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 51352 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 51353 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0838, + "step": 51354 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 51355 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0814, + "step": 51356 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 51357 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0821, + "step": 51358 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0678, + "step": 51359 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 51360 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 51361 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0805, + "step": 51362 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0878, + "step": 51363 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 51364 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 51365 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 51366 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 51367 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 51368 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 51369 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0901, + "step": 51370 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 51371 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 51372 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 51373 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 51374 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 51375 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 51376 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.091, + "step": 51377 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0802, + "step": 51378 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 51379 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 51380 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 51381 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 51382 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0838, + "step": 51383 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 51384 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 51385 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 51386 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 51387 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 51388 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 51389 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 51390 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 51391 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0859, + "step": 51392 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0805, + "step": 51393 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 51394 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0814, + "step": 51395 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0839, + "step": 51396 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 51397 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0835, + "step": 51398 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 51399 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 51400 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 51401 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 51402 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 51403 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 51404 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0795, + "step": 51405 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0752, + "step": 51406 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 51407 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0687, + "step": 51408 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 51409 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 51410 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.075, + "step": 51411 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 51412 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0743, + "step": 51413 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 51414 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0791, + "step": 51415 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 51416 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 51417 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 51418 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 51419 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 51420 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 51421 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 51422 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 51423 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 51424 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 51425 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 51426 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0865, + "step": 51427 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 51428 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0727, + "step": 51429 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 51430 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0765, + "step": 51431 + }, + { + "epoch": 3.12, + "learning_rate": 2e-05, + "loss": 0.0932, + "step": 51432 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 51433 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 51434 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 51435 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 51436 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 51437 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 51438 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 51439 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 51440 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 51441 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 51442 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 51443 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 51444 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 51445 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0824, + "step": 51446 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 51447 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 51448 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.091, + "step": 51449 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0967, + "step": 51450 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 51451 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 51452 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0847, + "step": 51453 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0842, + "step": 51454 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 51455 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 51456 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 51457 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0806, + "step": 51458 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 51459 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 51460 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0793, + "step": 51461 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 51462 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0883, + "step": 51463 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 51464 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 51465 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 51466 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 51467 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0898, + "step": 51468 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 51469 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 51470 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 51471 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 51472 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0967, + "step": 51473 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0916, + "step": 51474 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 51475 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 51476 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0845, + "step": 51477 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0771, + "step": 51478 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 51479 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.07, + "step": 51480 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0888, + "step": 51481 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 51482 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.088, + "step": 51483 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0941, + "step": 51484 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 51485 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0818, + "step": 51486 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 51487 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0737, + "step": 51488 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 51489 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 51490 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0789, + "step": 51491 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0793, + "step": 51492 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 51493 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0785, + "step": 51494 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0738, + "step": 51495 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 51496 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 51497 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1375, + "step": 51498 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 51499 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 51500 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 51501 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 51502 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0841, + "step": 51503 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 51504 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 51505 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 51506 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 51507 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0698, + "step": 51508 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0745, + "step": 51509 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 51510 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 51511 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 51512 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1446, + "step": 51513 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 51514 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 51515 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 51516 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 51517 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.091, + "step": 51518 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0911, + "step": 51519 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 51520 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 51521 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 51522 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 51523 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.082, + "step": 51524 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 51525 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0779, + "step": 51526 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 51527 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0786, + "step": 51528 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0804, + "step": 51529 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 51530 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 51531 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 51532 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 51533 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0803, + "step": 51534 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 51535 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 51536 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 51537 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0779, + "step": 51538 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 51539 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0914, + "step": 51540 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0832, + "step": 51541 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 51542 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 51543 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.088, + "step": 51544 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.082, + "step": 51545 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0897, + "step": 51546 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 51547 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 51548 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 51549 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0851, + "step": 51550 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 51551 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 51552 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0935, + "step": 51553 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 51554 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 51555 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 51556 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0718, + "step": 51557 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 51558 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0879, + "step": 51559 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 51560 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0713, + "step": 51561 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 51562 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0899, + "step": 51563 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 51564 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 51565 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 51566 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 51567 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 51568 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 51569 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 51570 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 51571 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 51572 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0781, + "step": 51573 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 51574 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 51575 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 51576 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0917, + "step": 51577 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 51578 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 51579 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0935, + "step": 51580 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0882, + "step": 51581 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0906, + "step": 51582 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 51583 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 51584 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 51585 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 51586 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0814, + "step": 51587 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 51588 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0842, + "step": 51589 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 51590 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 51591 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 51592 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 51593 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 51594 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.084, + "step": 51595 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 51596 + }, + { + "epoch": 3.13, + "learning_rate": 2e-05, + "loss": 0.0859, + "step": 51597 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0848, + "step": 51598 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0834, + "step": 51599 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0643, + "step": 51600 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 51601 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0967, + "step": 51602 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 51603 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 51604 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 51605 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 51606 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 51607 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0812, + "step": 51608 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 51609 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 51610 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 51611 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 51612 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 51613 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 51614 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 51615 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 51616 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 51617 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 51618 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0869, + "step": 51619 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 51620 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0887, + "step": 51621 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 51622 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 51623 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 51624 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 51625 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0807, + "step": 51626 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 51627 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 51628 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 51629 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 51630 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0852, + "step": 51631 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0807, + "step": 51632 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 51633 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0828, + "step": 51634 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 51635 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 51636 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 51637 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 51638 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 51639 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 51640 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 51641 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0796, + "step": 51642 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 51643 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 51644 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0834, + "step": 51645 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 51646 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 51647 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 51648 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0769, + "step": 51649 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 51650 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 51651 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0778, + "step": 51652 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 51653 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 51654 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 51655 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0935, + "step": 51656 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 51657 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 51658 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 51659 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.081, + "step": 51660 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 51661 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 51662 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 51663 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 51664 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 51665 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 51666 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.068, + "step": 51667 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 51668 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0874, + "step": 51669 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0749, + "step": 51670 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0812, + "step": 51671 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 51672 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 51673 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 51674 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0909, + "step": 51675 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 51676 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0683, + "step": 51677 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 51678 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0822, + "step": 51679 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0823, + "step": 51680 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 51681 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 51682 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0835, + "step": 51683 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 51684 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0694, + "step": 51685 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.077, + "step": 51686 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 51687 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 51688 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 51689 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 51690 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0732, + "step": 51691 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1465, + "step": 51692 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0838, + "step": 51693 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 51694 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 51695 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 51696 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 51697 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 51698 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 51699 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 51700 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0768, + "step": 51701 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 51702 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 51703 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 51704 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 51705 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1482, + "step": 51706 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0876, + "step": 51707 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 51708 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 51709 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 51710 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 51711 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 51712 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 51713 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 51714 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 51715 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 51716 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0845, + "step": 51717 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 51718 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0846, + "step": 51719 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 51720 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 51721 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0888, + "step": 51722 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0909, + "step": 51723 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 51724 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0771, + "step": 51725 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 51726 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 51727 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.082, + "step": 51728 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 51729 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 51730 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 51731 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 51732 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 51733 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 51734 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 51735 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0889, + "step": 51736 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0886, + "step": 51737 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 51738 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0894, + "step": 51739 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0816, + "step": 51740 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0875, + "step": 51741 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0898, + "step": 51742 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.073, + "step": 51743 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 51744 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0886, + "step": 51745 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0766, + "step": 51746 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 51747 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 51748 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 51749 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0881, + "step": 51750 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 51751 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0856, + "step": 51752 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.075, + "step": 51753 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 51754 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 51755 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0813, + "step": 51756 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0827, + "step": 51757 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 51758 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0701, + "step": 51759 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.0874, + "step": 51760 + }, + { + "epoch": 3.14, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 51761 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 51762 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0889, + "step": 51763 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 51764 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0812, + "step": 51765 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 51766 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 51767 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 51768 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0757, + "step": 51769 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 51770 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 51771 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.089, + "step": 51772 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0871, + "step": 51773 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 51774 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 51775 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 51776 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0898, + "step": 51777 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 51778 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 51779 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 51780 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 51781 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 51782 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0882, + "step": 51783 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0817, + "step": 51784 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 51785 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 51786 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 51787 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 51788 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0811, + "step": 51789 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0906, + "step": 51790 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 51791 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 51792 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0795, + "step": 51793 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 51794 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 51795 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0916, + "step": 51796 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 51797 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 51798 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0803, + "step": 51799 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 51800 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 51801 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 51802 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 51803 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 51804 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0911, + "step": 51805 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0829, + "step": 51806 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 51807 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 51808 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 51809 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 51810 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 51811 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 51812 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0865, + "step": 51813 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 51814 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0853, + "step": 51815 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 51816 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0918, + "step": 51817 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0692, + "step": 51818 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 51819 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0866, + "step": 51820 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 51821 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 51822 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 51823 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 51824 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 51825 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0906, + "step": 51826 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 51827 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0725, + "step": 51828 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0915, + "step": 51829 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 51830 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 51831 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 51832 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 51833 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 51834 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 51835 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 51836 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 51837 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 51838 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 51839 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 51840 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 51841 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 51842 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0853, + "step": 51843 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0856, + "step": 51844 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.085, + "step": 51845 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0903, + "step": 51846 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.068, + "step": 51847 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0918, + "step": 51848 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0783, + "step": 51849 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0878, + "step": 51850 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 51851 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 51852 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 51853 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 51854 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 51855 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 51856 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 51857 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 51858 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.089, + "step": 51859 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 51860 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 51861 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 51862 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 51863 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 51864 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 51865 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0842, + "step": 51866 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0838, + "step": 51867 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0837, + "step": 51868 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 51869 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 51870 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0821, + "step": 51871 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0852, + "step": 51872 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 51873 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0873, + "step": 51874 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 51875 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 51876 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0943, + "step": 51877 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 51878 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 51879 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 51880 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0839, + "step": 51881 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 51882 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 51883 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0869, + "step": 51884 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0888, + "step": 51885 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 51886 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 51887 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 51888 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 51889 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.066, + "step": 51890 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0857, + "step": 51891 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 51892 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0616, + "step": 51893 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 51894 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 51895 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0911, + "step": 51896 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.079, + "step": 51897 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 51898 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0871, + "step": 51899 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 51900 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 51901 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 51902 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 51903 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0857, + "step": 51904 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 51905 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 51906 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 51907 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0943, + "step": 51908 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 51909 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 51910 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 51911 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 51912 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 51913 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.091, + "step": 51914 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 51915 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 51916 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 51917 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0897, + "step": 51918 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 51919 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 51920 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 51921 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0909, + "step": 51922 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 51923 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 51924 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 51925 + }, + { + "epoch": 3.15, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 51926 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 51927 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 51928 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 51929 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 51930 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 51931 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 51932 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0851, + "step": 51933 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 51934 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0914, + "step": 51935 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 51936 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.078, + "step": 51937 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 51938 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0846, + "step": 51939 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 51940 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0873, + "step": 51941 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0727, + "step": 51942 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 51943 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0826, + "step": 51944 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 51945 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 51946 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0643, + "step": 51947 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 51948 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 51949 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 51950 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 51951 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0847, + "step": 51952 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0724, + "step": 51953 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0952, + "step": 51954 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 51955 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 51956 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 51957 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 51958 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 51959 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 51960 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 51961 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 51962 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 51963 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0892, + "step": 51964 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 51965 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 51966 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 51967 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.082, + "step": 51968 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 51969 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0854, + "step": 51970 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 51971 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 51972 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 51973 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.077, + "step": 51974 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 51975 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 51976 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0825, + "step": 51977 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 51978 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0935, + "step": 51979 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 51980 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 51981 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 51982 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 51983 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0866, + "step": 51984 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 51985 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 51986 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0796, + "step": 51987 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 51988 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0821, + "step": 51989 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 51990 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 51991 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 51992 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0902, + "step": 51993 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 51994 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0868, + "step": 51995 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 51996 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 51997 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0867, + "step": 51998 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 51999 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0786, + "step": 52000 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 52001 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 52002 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 52003 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 52004 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 52005 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 52006 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0901, + "step": 52007 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0967, + "step": 52008 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 52009 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 52010 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0804, + "step": 52011 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 52012 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0876, + "step": 52013 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 52014 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 52015 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 52016 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0748, + "step": 52017 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0883, + "step": 52018 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 52019 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0794, + "step": 52020 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0773, + "step": 52021 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 52022 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 52023 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 52024 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 52025 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 52026 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 52027 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 52028 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0775, + "step": 52029 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 52030 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0778, + "step": 52031 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 52032 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 52033 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0611, + "step": 52034 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0765, + "step": 52035 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 52036 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 52037 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 52038 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 52039 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 52040 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 52041 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0906, + "step": 52042 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 52043 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 52044 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 52045 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 52046 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 52047 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0735, + "step": 52048 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 52049 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 52050 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 52051 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 52052 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 52053 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 52054 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0894, + "step": 52055 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 52056 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 52057 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 52058 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 52059 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 52060 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 52061 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 52062 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.09, + "step": 52063 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 52064 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 52065 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0857, + "step": 52066 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 52067 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 52068 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 52069 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0837, + "step": 52070 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 52071 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1477, + "step": 52072 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1319, + "step": 52073 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 52074 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 52075 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0815, + "step": 52076 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 52077 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 52078 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 52079 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0839, + "step": 52080 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 52081 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 52082 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0862, + "step": 52083 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 52084 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 52085 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 52086 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 52087 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0862, + "step": 52088 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 52089 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 52090 + }, + { + "epoch": 3.16, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 52091 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0879, + "step": 52092 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0741, + "step": 52093 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 52094 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 52095 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 52096 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0902, + "step": 52097 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.082, + "step": 52098 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0837, + "step": 52099 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 52100 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 52101 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 52102 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 52103 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 52104 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 52105 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 52106 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 52107 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 52108 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 52109 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 52110 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 52111 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 52112 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 52113 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 52114 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 52115 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 52116 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0797, + "step": 52117 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0785, + "step": 52118 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0771, + "step": 52119 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0874, + "step": 52120 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0889, + "step": 52121 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0806, + "step": 52122 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 52123 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 52124 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 52125 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 52126 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0832, + "step": 52127 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0848, + "step": 52128 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 52129 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 52130 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 52131 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 52132 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0881, + "step": 52133 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0881, + "step": 52134 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 52135 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 52136 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 52137 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 52138 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 52139 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0785, + "step": 52140 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 52141 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 52142 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0842, + "step": 52143 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 52144 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 52145 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 52146 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 52147 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 52148 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 52149 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0843, + "step": 52150 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 52151 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 52152 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 52153 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0817, + "step": 52154 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 52155 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 52156 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 52157 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0941, + "step": 52158 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 52159 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 52160 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 52161 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0868, + "step": 52162 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 52163 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 52164 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 52165 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 52166 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0864, + "step": 52167 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 52168 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0698, + "step": 52169 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 52170 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 52171 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.16, + "step": 52172 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 52173 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 52174 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 52175 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 52176 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.081, + "step": 52177 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 52178 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 52179 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 52180 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 52181 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 52182 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 52183 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 52184 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0789, + "step": 52185 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0868, + "step": 52186 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0797, + "step": 52187 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0886, + "step": 52188 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0895, + "step": 52189 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 52190 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 52191 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0699, + "step": 52192 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 52193 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0887, + "step": 52194 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 52195 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0803, + "step": 52196 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 52197 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 52198 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 52199 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 52200 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0895, + "step": 52201 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 52202 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0855, + "step": 52203 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 52204 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 52205 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0807, + "step": 52206 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 52207 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 52208 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 52209 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 52210 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0873, + "step": 52211 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 52212 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0864, + "step": 52213 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0863, + "step": 52214 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 52215 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 52216 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 52217 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 52218 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0863, + "step": 52219 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0796, + "step": 52220 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 52221 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 52222 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 52223 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 52224 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 52225 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.088, + "step": 52226 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0777, + "step": 52227 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 52228 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 52229 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1549, + "step": 52230 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 52231 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 52232 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 52233 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 52234 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0898, + "step": 52235 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0862, + "step": 52236 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.087, + "step": 52237 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 52238 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0625, + "step": 52239 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.078, + "step": 52240 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 52241 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 52242 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 52243 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 52244 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0858, + "step": 52245 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0809, + "step": 52246 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 52247 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 52248 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 52249 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 52250 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 52251 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.0872, + "step": 52252 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.084, + "step": 52253 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.084, + "step": 52254 + }, + { + "epoch": 3.17, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 52255 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 52256 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 52257 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 52258 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 52259 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 52260 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 52261 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 52262 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 52263 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 52264 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0761, + "step": 52265 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 52266 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0766, + "step": 52267 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 52268 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 52269 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 52270 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0868, + "step": 52271 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 52272 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0743, + "step": 52273 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0842, + "step": 52274 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 52275 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 52276 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 52277 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0903, + "step": 52278 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0881, + "step": 52279 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 52280 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0796, + "step": 52281 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 52282 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 52283 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 52284 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 52285 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 52286 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0888, + "step": 52287 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 52288 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 52289 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0792, + "step": 52290 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 52291 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0875, + "step": 52292 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 52293 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 52294 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 52295 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0848, + "step": 52296 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 52297 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 52298 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 52299 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0663, + "step": 52300 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 52301 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 52302 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 52303 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0813, + "step": 52304 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0883, + "step": 52305 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 52306 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 52307 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 52308 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 52309 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 52310 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 52311 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0834, + "step": 52312 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 52313 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 52314 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 52315 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 52316 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 52317 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0653, + "step": 52318 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 52319 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 52320 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 52321 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.085, + "step": 52322 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1565, + "step": 52323 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 52324 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0832, + "step": 52325 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 52326 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 52327 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0877, + "step": 52328 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 52329 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 52330 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0897, + "step": 52331 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 52332 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0897, + "step": 52333 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 52334 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 52335 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 52336 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0929, + "step": 52337 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 52338 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 52339 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 52340 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 52341 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 52342 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 52343 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 52344 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 52345 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 52346 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0888, + "step": 52347 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1504, + "step": 52348 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 52349 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 52350 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0842, + "step": 52351 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 52352 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 52353 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 52354 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0819, + "step": 52355 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 52356 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 52357 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 52358 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.084, + "step": 52359 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 52360 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 52361 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 52362 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 52363 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 52364 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0903, + "step": 52365 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 52366 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 52367 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 52368 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 52369 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 52370 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0914, + "step": 52371 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0735, + "step": 52372 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0943, + "step": 52373 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 52374 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 52375 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 52376 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 52377 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0795, + "step": 52378 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 52379 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 52380 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 52381 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0813, + "step": 52382 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 52383 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0753, + "step": 52384 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0881, + "step": 52385 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0902, + "step": 52386 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 52387 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 52388 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 52389 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 52390 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 52391 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 52392 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 52393 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 52394 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 52395 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.091, + "step": 52396 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 52397 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 52398 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0851, + "step": 52399 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 52400 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1529, + "step": 52401 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 52402 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0777, + "step": 52403 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 52404 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 52405 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 52406 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0967, + "step": 52407 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 52408 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.082, + "step": 52409 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.091, + "step": 52410 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0772, + "step": 52411 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 52412 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.075, + "step": 52413 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 52414 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 52415 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0911, + "step": 52416 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 52417 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0668, + "step": 52418 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0889, + "step": 52419 + }, + { + "epoch": 3.18, + "learning_rate": 2e-05, + "loss": 0.0895, + "step": 52420 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 52421 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 52422 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 52423 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 52424 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 52425 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 52426 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 52427 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 52428 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0729, + "step": 52429 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 52430 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 52431 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 52432 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 52433 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 52434 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 52435 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 52436 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 52437 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 52438 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 52439 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0838, + "step": 52440 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 52441 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 52442 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 52443 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 52444 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 52445 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 52446 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 52447 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.078, + "step": 52448 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 52449 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0847, + "step": 52450 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 52451 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 52452 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0714, + "step": 52453 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 52454 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 52455 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0941, + "step": 52456 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 52457 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 52458 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 52459 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 52460 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0743, + "step": 52461 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0695, + "step": 52462 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0752, + "step": 52463 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 52464 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0771, + "step": 52465 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0749, + "step": 52466 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0832, + "step": 52467 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0747, + "step": 52468 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.09, + "step": 52469 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 52470 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 52471 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0831, + "step": 52472 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 52473 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 52474 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0867, + "step": 52475 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.081, + "step": 52476 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 52477 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0877, + "step": 52478 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 52479 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0799, + "step": 52480 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 52481 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 52482 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 52483 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 52484 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 52485 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 52486 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 52487 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 52488 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 52489 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 52490 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0823, + "step": 52491 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0849, + "step": 52492 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 52493 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 52494 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1457, + "step": 52495 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 52496 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 52497 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 52498 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 52499 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0848, + "step": 52500 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0867, + "step": 52501 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 52502 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0855, + "step": 52503 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 52504 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 52505 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 52506 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0793, + "step": 52507 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 52508 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 52509 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0779, + "step": 52510 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 52511 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 52512 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 52513 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0894, + "step": 52514 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0851, + "step": 52515 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 52516 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0847, + "step": 52517 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 52518 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 52519 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 52520 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 52521 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 52522 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 52523 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0862, + "step": 52524 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0807, + "step": 52525 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 52526 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 52527 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 52528 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0891, + "step": 52529 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 52530 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0867, + "step": 52531 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 52532 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 52533 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 52534 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 52535 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0785, + "step": 52536 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.082, + "step": 52537 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 52538 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 52539 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 52540 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 52541 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 52542 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 52543 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 52544 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 52545 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 52546 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0806, + "step": 52547 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 52548 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 52549 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 52550 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 52551 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0714, + "step": 52552 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 52553 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 52554 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0804, + "step": 52555 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 52556 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.075, + "step": 52557 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 52558 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 52559 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 52560 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 52561 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 52562 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0832, + "step": 52563 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 52564 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0898, + "step": 52565 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 52566 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0853, + "step": 52567 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 52568 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 52569 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 52570 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0667, + "step": 52571 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 52572 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0759, + "step": 52573 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0859, + "step": 52574 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 52575 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0845, + "step": 52576 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 52577 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 52578 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.073, + "step": 52579 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0874, + "step": 52580 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 52581 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0809, + "step": 52582 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0756, + "step": 52583 + }, + { + "epoch": 3.19, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 52584 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0876, + "step": 52585 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 52586 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0899, + "step": 52587 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0872, + "step": 52588 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0843, + "step": 52589 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0909, + "step": 52590 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.088, + "step": 52591 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 52592 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0795, + "step": 52593 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 52594 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0851, + "step": 52595 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 52596 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0935, + "step": 52597 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 52598 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 52599 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 52600 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 52601 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 52602 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0879, + "step": 52603 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0798, + "step": 52604 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.09, + "step": 52605 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 52606 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 52607 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.091, + "step": 52608 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 52609 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0853, + "step": 52610 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 52611 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 52612 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0866, + "step": 52613 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 52614 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 52615 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 52616 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0856, + "step": 52617 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0635, + "step": 52618 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0806, + "step": 52619 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 52620 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 52621 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0805, + "step": 52622 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 52623 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 52624 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 52625 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 52626 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 52627 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0866, + "step": 52628 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0849, + "step": 52629 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 52630 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0894, + "step": 52631 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 52632 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 52633 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 52634 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 52635 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0793, + "step": 52636 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 52637 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 52638 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 52639 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0789, + "step": 52640 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 52641 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 52642 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0911, + "step": 52643 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 52644 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 52645 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 52646 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 52647 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 52648 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 52649 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 52650 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0867, + "step": 52651 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0807, + "step": 52652 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 52653 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0849, + "step": 52654 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 52655 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0774, + "step": 52656 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0812, + "step": 52657 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0852, + "step": 52658 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0826, + "step": 52659 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 52660 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 52661 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 52662 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 52663 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 52664 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 52665 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0782, + "step": 52666 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 52667 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 52668 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 52669 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0837, + "step": 52670 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 52671 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0788, + "step": 52672 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0889, + "step": 52673 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 52674 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0863, + "step": 52675 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0827, + "step": 52676 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 52677 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 52678 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0846, + "step": 52679 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0832, + "step": 52680 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 52681 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0901, + "step": 52682 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 52683 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 52684 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 52685 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0791, + "step": 52686 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 52687 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 52688 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 52689 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 52690 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0763, + "step": 52691 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 52692 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0891, + "step": 52693 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 52694 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 52695 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0796, + "step": 52696 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 52697 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0857, + "step": 52698 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 52699 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 52700 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 52701 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0873, + "step": 52702 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 52703 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 52704 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0929, + "step": 52705 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 52706 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 52707 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 52708 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0828, + "step": 52709 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 52710 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 52711 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 52712 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0862, + "step": 52713 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0787, + "step": 52714 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 52715 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 52716 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0889, + "step": 52717 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 52718 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 52719 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0713, + "step": 52720 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 52721 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 52722 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0833, + "step": 52723 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 52724 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 52725 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0911, + "step": 52726 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 52727 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 52728 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 52729 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 52730 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0875, + "step": 52731 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 52732 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0901, + "step": 52733 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0707, + "step": 52734 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 52735 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0864, + "step": 52736 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0849, + "step": 52737 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 52738 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0833, + "step": 52739 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 52740 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 52741 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 52742 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 52743 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 52744 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 52745 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 52746 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 52747 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 52748 + }, + { + "epoch": 3.2, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 52749 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0784, + "step": 52750 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0863, + "step": 52751 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 52752 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0869, + "step": 52753 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 52754 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0745, + "step": 52755 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 52756 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0787, + "step": 52757 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 52758 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 52759 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 52760 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 52761 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0932, + "step": 52762 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 52763 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0868, + "step": 52764 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 52765 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0787, + "step": 52766 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 52767 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 52768 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.077, + "step": 52769 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0824, + "step": 52770 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0876, + "step": 52771 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0858, + "step": 52772 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 52773 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0781, + "step": 52774 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0929, + "step": 52775 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 52776 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 52777 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 52778 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0806, + "step": 52779 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 52780 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 52781 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 52782 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 52783 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 52784 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 52785 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0906, + "step": 52786 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 52787 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 52788 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 52789 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0911, + "step": 52790 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 52791 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0903, + "step": 52792 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 52793 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 52794 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 52795 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 52796 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 52797 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0794, + "step": 52798 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 52799 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 52800 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 52801 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 52802 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.079, + "step": 52803 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 52804 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 52805 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 52806 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 52807 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 52808 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 52809 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 52810 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 52811 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.089, + "step": 52812 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0754, + "step": 52813 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0686, + "step": 52814 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 52815 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 52816 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 52817 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 52818 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0659, + "step": 52819 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0903, + "step": 52820 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 52821 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 52822 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 52823 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 52824 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0858, + "step": 52825 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 52826 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0903, + "step": 52827 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0815, + "step": 52828 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 52829 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0875, + "step": 52830 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 52831 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 52832 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 52833 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 52834 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 52835 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 52836 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0855, + "step": 52837 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 52838 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0878, + "step": 52839 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0733, + "step": 52840 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 52841 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0726, + "step": 52842 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 52843 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 52844 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0664, + "step": 52845 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 52846 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 52847 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 52848 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0802, + "step": 52849 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 52850 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 52851 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 52852 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1465, + "step": 52853 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 52854 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 52855 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0802, + "step": 52856 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 52857 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 52858 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 52859 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0847, + "step": 52860 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0788, + "step": 52861 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1426, + "step": 52862 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 52863 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 52864 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 52865 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 52866 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 52867 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0883, + "step": 52868 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 52869 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 52870 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0911, + "step": 52871 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 52872 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0814, + "step": 52873 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0823, + "step": 52874 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 52875 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 52876 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 52877 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1328, + "step": 52878 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 52879 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 52880 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 52881 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 52882 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 52883 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 52884 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 52885 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 52886 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 52887 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0751, + "step": 52888 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 52889 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 52890 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 52891 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 52892 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 52893 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 52894 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 52895 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.085, + "step": 52896 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 52897 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 52898 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 52899 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0788, + "step": 52900 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 52901 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0826, + "step": 52902 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 52903 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0911, + "step": 52904 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 52905 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 52906 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 52907 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 52908 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 52909 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0867, + "step": 52910 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 52911 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0857, + "step": 52912 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 52913 + }, + { + "epoch": 3.21, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 52914 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 52915 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 52916 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 52917 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 52918 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0914, + "step": 52919 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 52920 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 52921 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 52922 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 52923 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 52924 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 52925 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 52926 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 52927 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0872, + "step": 52928 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 52929 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 52930 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 52931 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 52932 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 52933 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 52934 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0822, + "step": 52935 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0909, + "step": 52936 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.08, + "step": 52937 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0811, + "step": 52938 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 52939 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 52940 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0664, + "step": 52941 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0827, + "step": 52942 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 52943 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 52944 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 52945 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0827, + "step": 52946 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 52947 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 52948 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 52949 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1325, + "step": 52950 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 52951 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 52952 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 52953 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0911, + "step": 52954 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 52955 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 52956 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1242, + "step": 52957 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0903, + "step": 52958 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 52959 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 52960 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 52961 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0655, + "step": 52962 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.083, + "step": 52963 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 52964 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0852, + "step": 52965 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0943, + "step": 52966 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0843, + "step": 52967 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.083, + "step": 52968 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 52969 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 52970 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0941, + "step": 52971 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 52972 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 52973 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0834, + "step": 52974 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 52975 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 52976 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0746, + "step": 52977 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 52978 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 52979 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 52980 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0774, + "step": 52981 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0696, + "step": 52982 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0887, + "step": 52983 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0762, + "step": 52984 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 52985 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 52986 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 52987 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 52988 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0772, + "step": 52989 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0786, + "step": 52990 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 52991 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0874, + "step": 52992 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 52993 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 52994 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 52995 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0856, + "step": 52996 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0874, + "step": 52997 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0891, + "step": 52998 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0802, + "step": 52999 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 53000 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 53001 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 53002 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 53003 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 53004 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 53005 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 53006 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 53007 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 53008 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 53009 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 53010 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0892, + "step": 53011 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 53012 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0857, + "step": 53013 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0736, + "step": 53014 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 53015 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 53016 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 53017 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 53018 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 53019 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 53020 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 53021 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 53022 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.085, + "step": 53023 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 53024 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 53025 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0792, + "step": 53026 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 53027 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0838, + "step": 53028 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 53029 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 53030 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 53031 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.089, + "step": 53032 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 53033 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 53034 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.081, + "step": 53035 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 53036 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 53037 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 53038 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 53039 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 53040 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 53041 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 53042 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 53043 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 53044 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 53045 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 53046 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 53047 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 53048 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 53049 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.079, + "step": 53050 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 53051 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0819, + "step": 53052 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 53053 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1373, + "step": 53054 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0856, + "step": 53055 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 53056 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 53057 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0816, + "step": 53058 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 53059 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 53060 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 53061 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 53062 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 53063 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 53064 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 53065 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 53066 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0833, + "step": 53067 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 53068 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 53069 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 53070 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 53071 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 53072 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0894, + "step": 53073 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 53074 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 53075 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 53076 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 53077 + }, + { + "epoch": 3.22, + "learning_rate": 2e-05, + "loss": 0.0771, + "step": 53078 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 53079 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0917, + "step": 53080 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0875, + "step": 53081 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 53082 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 53083 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 53084 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 53085 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 53086 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 53087 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 53088 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 53089 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 53090 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 53091 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 53092 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 53093 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 53094 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 53095 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0728, + "step": 53096 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0906, + "step": 53097 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0801, + "step": 53098 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 53099 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 53100 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0792, + "step": 53101 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 53102 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 53103 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 53104 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 53105 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0776, + "step": 53106 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 53107 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0736, + "step": 53108 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 53109 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 53110 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0844, + "step": 53111 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 53112 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 53113 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0901, + "step": 53114 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0679, + "step": 53115 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 53116 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 53117 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 53118 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 53119 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 53120 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.081, + "step": 53121 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 53122 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 53123 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 53124 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 53125 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 53126 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 53127 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 53128 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 53129 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 53130 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 53131 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 53132 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0772, + "step": 53133 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 53134 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0879, + "step": 53135 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 53136 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 53137 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 53138 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0858, + "step": 53139 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 53140 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 53141 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0711, + "step": 53142 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0881, + "step": 53143 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 53144 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 53145 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 53146 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 53147 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0855, + "step": 53148 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 53149 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 53150 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 53151 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0846, + "step": 53152 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 53153 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0838, + "step": 53154 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 53155 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 53156 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 53157 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0943, + "step": 53158 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0872, + "step": 53159 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 53160 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 53161 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 53162 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 53163 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 53164 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.09, + "step": 53165 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 53166 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.087, + "step": 53167 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0845, + "step": 53168 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 53169 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0865, + "step": 53170 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 53171 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0856, + "step": 53172 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0798, + "step": 53173 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 53174 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 53175 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0874, + "step": 53176 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 53177 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 53178 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0794, + "step": 53179 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 53180 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 53181 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 53182 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 53183 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 53184 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.082, + "step": 53185 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 53186 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 53187 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 53188 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 53189 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0804, + "step": 53190 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 53191 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 53192 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 53193 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 53194 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0823, + "step": 53195 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0935, + "step": 53196 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1295, + "step": 53197 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 53198 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 53199 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 53200 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 53201 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0943, + "step": 53202 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 53203 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0821, + "step": 53204 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0899, + "step": 53205 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 53206 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0844, + "step": 53207 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0808, + "step": 53208 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0801, + "step": 53209 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 53210 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 53211 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.09, + "step": 53212 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0903, + "step": 53213 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 53214 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 53215 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 53216 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.085, + "step": 53217 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0749, + "step": 53218 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 53219 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0847, + "step": 53220 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 53221 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 53222 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 53223 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0876, + "step": 53224 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0888, + "step": 53225 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 53226 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 53227 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 53228 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 53229 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0725, + "step": 53230 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 53231 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 53232 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0895, + "step": 53233 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 53234 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 53235 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0779, + "step": 53236 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0782, + "step": 53237 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0739, + "step": 53238 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 53239 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0935, + "step": 53240 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 53241 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0936, + "step": 53242 + }, + { + "epoch": 3.23, + "learning_rate": 2e-05, + "loss": 0.0771, + "step": 53243 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 53244 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0639, + "step": 53245 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 53246 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 53247 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 53248 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 53249 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 53250 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.082, + "step": 53251 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 53252 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 53253 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0867, + "step": 53254 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 53255 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 53256 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 53257 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0756, + "step": 53258 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0808, + "step": 53259 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 53260 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 53261 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 53262 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 53263 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 53264 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 53265 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 53266 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0789, + "step": 53267 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 53268 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 53269 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0842, + "step": 53270 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 53271 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 53272 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0819, + "step": 53273 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.066, + "step": 53274 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0902, + "step": 53275 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 53276 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0694, + "step": 53277 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 53278 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 53279 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 53280 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 53281 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 53282 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 53283 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 53284 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0776, + "step": 53285 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0834, + "step": 53286 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0916, + "step": 53287 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 53288 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 53289 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0903, + "step": 53290 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 53291 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 53292 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 53293 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 53294 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 53295 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 53296 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 53297 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 53298 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 53299 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0878, + "step": 53300 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0819, + "step": 53301 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 53302 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 53303 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 53304 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 53305 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 53306 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 53307 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0888, + "step": 53308 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 53309 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0849, + "step": 53310 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 53311 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0894, + "step": 53312 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 53313 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0899, + "step": 53314 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0748, + "step": 53315 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 53316 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 53317 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 53318 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0877, + "step": 53319 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 53320 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1455, + "step": 53321 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 53322 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0863, + "step": 53323 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 53324 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 53325 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 53326 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0803, + "step": 53327 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 53328 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 53329 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0873, + "step": 53330 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 53331 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 53332 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0886, + "step": 53333 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0799, + "step": 53334 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 53335 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 53336 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0752, + "step": 53337 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0833, + "step": 53338 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 53339 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 53340 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0899, + "step": 53341 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 53342 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 53343 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 53344 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 53345 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 53346 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 53347 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 53348 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 53349 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0929, + "step": 53350 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0906, + "step": 53351 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 53352 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 53353 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 53354 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0801, + "step": 53355 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 53356 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0821, + "step": 53357 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 53358 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 53359 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 53360 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 53361 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 53362 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 53363 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0859, + "step": 53364 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 53365 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 53366 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 53367 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 53368 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 53369 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 53370 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 53371 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0879, + "step": 53372 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0843, + "step": 53373 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 53374 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0891, + "step": 53375 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 53376 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 53377 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 53378 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 53379 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 53380 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 53381 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0701, + "step": 53382 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 53383 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 53384 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0771, + "step": 53385 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 53386 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0797, + "step": 53387 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0918, + "step": 53388 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0734, + "step": 53389 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0828, + "step": 53390 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 53391 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 53392 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 53393 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0879, + "step": 53394 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 53395 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0891, + "step": 53396 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 53397 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 53398 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1312, + "step": 53399 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 53400 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 53401 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 53402 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 53403 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 53404 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 53405 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 53406 + }, + { + "epoch": 3.24, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 53407 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 53408 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 53409 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 53410 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 53411 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 53412 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 53413 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 53414 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0805, + "step": 53415 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 53416 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 53417 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 53418 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 53419 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.084, + "step": 53420 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 53421 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 53422 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 53423 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 53424 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0815, + "step": 53425 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0895, + "step": 53426 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0916, + "step": 53427 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0735, + "step": 53428 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 53429 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 53430 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0784, + "step": 53431 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 53432 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 53433 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 53434 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1395, + "step": 53435 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 53436 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 53437 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.074, + "step": 53438 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 53439 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 53440 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 53441 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 53442 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 53443 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 53444 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.078, + "step": 53445 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0733, + "step": 53446 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0875, + "step": 53447 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 53448 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 53449 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 53450 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 53451 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 53452 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 53453 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 53454 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 53455 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0898, + "step": 53456 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 53457 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.14, + "step": 53458 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 53459 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 53460 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 53461 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 53462 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0815, + "step": 53463 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 53464 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 53465 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 53466 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 53467 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 53468 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 53469 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0849, + "step": 53470 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0742, + "step": 53471 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 53472 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.083, + "step": 53473 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 53474 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 53475 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 53476 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 53477 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 53478 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 53479 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0901, + "step": 53480 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 53481 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0827, + "step": 53482 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 53483 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0869, + "step": 53484 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 53485 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0837, + "step": 53486 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 53487 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 53488 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 53489 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 53490 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 53491 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 53492 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0807, + "step": 53493 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 53494 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 53495 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 53496 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0851, + "step": 53497 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 53498 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 53499 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 53500 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 53501 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 53502 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0898, + "step": 53503 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 53504 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 53505 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 53506 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0792, + "step": 53507 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0831, + "step": 53508 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 53509 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 53510 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 53511 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0901, + "step": 53512 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0906, + "step": 53513 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 53514 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0823, + "step": 53515 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0873, + "step": 53516 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 53517 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0943, + "step": 53518 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0825, + "step": 53519 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0929, + "step": 53520 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0849, + "step": 53521 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 53522 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 53523 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 53524 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0731, + "step": 53525 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 53526 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 53527 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 53528 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 53529 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 53530 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0676, + "step": 53531 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 53532 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0854, + "step": 53533 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 53534 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 53535 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 53536 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 53537 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 53538 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 53539 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 53540 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 53541 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 53542 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 53543 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 53544 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 53545 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0854, + "step": 53546 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 53547 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 53548 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 53549 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0917, + "step": 53550 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0749, + "step": 53551 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0916, + "step": 53552 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 53553 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0803, + "step": 53554 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 53555 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 53556 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0865, + "step": 53557 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 53558 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0901, + "step": 53559 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0814, + "step": 53560 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 53561 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1597, + "step": 53562 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 53563 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 53564 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0762, + "step": 53565 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 53566 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 53567 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 53568 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 53569 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 53570 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0849, + "step": 53571 + }, + { + "epoch": 3.25, + "learning_rate": 2e-05, + "loss": 0.0941, + "step": 53572 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 53573 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 53574 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 53575 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 53576 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0898, + "step": 53577 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 53578 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0819, + "step": 53579 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 53580 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 53581 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0918, + "step": 53582 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0761, + "step": 53583 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 53584 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0844, + "step": 53585 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 53586 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 53587 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 53588 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 53589 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 53590 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 53591 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0748, + "step": 53592 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0871, + "step": 53593 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0797, + "step": 53594 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 53595 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0941, + "step": 53596 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 53597 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0841, + "step": 53598 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0895, + "step": 53599 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 53600 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 53601 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 53602 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 53603 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 53604 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 53605 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 53606 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 53607 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0866, + "step": 53608 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0845, + "step": 53609 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 53610 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0932, + "step": 53611 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 53612 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 53613 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 53614 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 53615 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1394, + "step": 53616 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 53617 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0798, + "step": 53618 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 53619 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 53620 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0943, + "step": 53621 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0886, + "step": 53622 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0791, + "step": 53623 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 53624 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0794, + "step": 53625 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 53626 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0864, + "step": 53627 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 53628 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 53629 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 53630 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 53631 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 53632 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0697, + "step": 53633 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0883, + "step": 53634 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 53635 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 53636 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 53637 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 53638 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 53639 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0775, + "step": 53640 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0707, + "step": 53641 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 53642 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 53643 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 53644 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 53645 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 53646 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 53647 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0879, + "step": 53648 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0847, + "step": 53649 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0886, + "step": 53650 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0773, + "step": 53651 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0898, + "step": 53652 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0874, + "step": 53653 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 53654 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 53655 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0842, + "step": 53656 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 53657 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 53658 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0705, + "step": 53659 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 53660 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 53661 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 53662 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 53663 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 53664 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0877, + "step": 53665 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0914, + "step": 53666 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 53667 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 53668 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0943, + "step": 53669 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 53670 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 53671 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 53672 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 53673 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 53674 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 53675 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 53676 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 53677 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0824, + "step": 53678 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 53679 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0842, + "step": 53680 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.073, + "step": 53681 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 53682 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 53683 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 53684 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0783, + "step": 53685 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 53686 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 53687 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 53688 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 53689 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 53690 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 53691 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 53692 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0918, + "step": 53693 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0887, + "step": 53694 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 53695 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 53696 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 53697 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 53698 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 53699 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 53700 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0909, + "step": 53701 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0873, + "step": 53702 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 53703 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0807, + "step": 53704 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 53705 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 53706 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 53707 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0797, + "step": 53708 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 53709 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 53710 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 53711 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 53712 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 53713 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 53714 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 53715 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 53716 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 53717 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 53718 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 53719 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 53720 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0872, + "step": 53721 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 53722 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 53723 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 53724 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0775, + "step": 53725 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 53726 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 53727 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 53728 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 53729 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0701, + "step": 53730 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 53731 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 53732 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 53733 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 53734 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 53735 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 53736 + }, + { + "epoch": 3.26, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 53737 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 53738 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 53739 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 53740 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 53741 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 53742 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 53743 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0751, + "step": 53744 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 53745 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 53746 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 53747 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0869, + "step": 53748 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 53749 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 53750 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 53751 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 53752 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 53753 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0895, + "step": 53754 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0935, + "step": 53755 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 53756 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0818, + "step": 53757 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0753, + "step": 53758 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 53759 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0906, + "step": 53760 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 53761 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0834, + "step": 53762 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 53763 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 53764 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 53765 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 53766 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 53767 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 53768 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 53769 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 53770 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 53771 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 53772 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0874, + "step": 53773 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 53774 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0732, + "step": 53775 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 53776 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 53777 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 53778 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0859, + "step": 53779 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 53780 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 53781 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 53782 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0869, + "step": 53783 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 53784 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 53785 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 53786 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 53787 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 53788 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 53789 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 53790 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0936, + "step": 53791 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 53792 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 53793 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0841, + "step": 53794 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 53795 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0932, + "step": 53796 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 53797 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 53798 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 53799 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 53800 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 53801 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0787, + "step": 53802 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 53803 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 53804 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 53805 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 53806 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 53807 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0817, + "step": 53808 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 53809 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 53810 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0818, + "step": 53811 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 53812 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.082, + "step": 53813 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 53814 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 53815 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 53816 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 53817 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 53818 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0834, + "step": 53819 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 53820 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0852, + "step": 53821 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0824, + "step": 53822 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 53823 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 53824 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0899, + "step": 53825 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 53826 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 53827 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 53828 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 53829 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0819, + "step": 53830 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 53831 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0886, + "step": 53832 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 53833 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 53834 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 53835 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 53836 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 53837 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0883, + "step": 53838 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0898, + "step": 53839 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0865, + "step": 53840 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 53841 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 53842 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0902, + "step": 53843 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0952, + "step": 53844 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0859, + "step": 53845 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0813, + "step": 53846 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0936, + "step": 53847 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 53848 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 53849 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0859, + "step": 53850 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 53851 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 53852 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 53853 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 53854 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0889, + "step": 53855 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 53856 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 53857 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 53858 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0855, + "step": 53859 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 53860 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 53861 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 53862 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0793, + "step": 53863 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 53864 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 53865 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 53866 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 53867 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0883, + "step": 53868 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 53869 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 53870 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.091, + "step": 53871 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0799, + "step": 53872 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 53873 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 53874 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 53875 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 53876 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 53877 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0737, + "step": 53878 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 53879 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 53880 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 53881 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0711, + "step": 53882 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 53883 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 53884 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 53885 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 53886 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 53887 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 53888 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 53889 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0806, + "step": 53890 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0801, + "step": 53891 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 53892 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0815, + "step": 53893 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0845, + "step": 53894 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 53895 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0725, + "step": 53896 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 53897 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 53898 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 53899 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 53900 + }, + { + "epoch": 3.27, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 53901 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 53902 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 53903 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 53904 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 53905 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 53906 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 53907 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 53908 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 53909 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 53910 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 53911 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 53912 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 53913 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0833, + "step": 53914 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 53915 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 53916 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0743, + "step": 53917 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 53918 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 53919 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 53920 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 53921 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0901, + "step": 53922 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 53923 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0827, + "step": 53924 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 53925 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 53926 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0863, + "step": 53927 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0685, + "step": 53928 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0874, + "step": 53929 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 53930 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 53931 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0892, + "step": 53932 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 53933 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 53934 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0941, + "step": 53935 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0843, + "step": 53936 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0863, + "step": 53937 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 53938 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 53939 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 53940 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 53941 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 53942 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 53943 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 53944 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 53945 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0836, + "step": 53946 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 53947 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0725, + "step": 53948 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 53949 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0772, + "step": 53950 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 53951 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 53952 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 53953 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 53954 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 53955 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 53956 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 53957 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 53958 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 53959 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 53960 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 53961 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 53962 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 53963 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0879, + "step": 53964 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 53965 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0822, + "step": 53966 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 53967 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 53968 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0851, + "step": 53969 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0788, + "step": 53970 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 53971 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0696, + "step": 53972 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 53973 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 53974 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 53975 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 53976 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 53977 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 53978 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 53979 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 53980 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 53981 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 53982 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 53983 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 53984 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 53985 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0814, + "step": 53986 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 53987 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 53988 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 53989 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 53990 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 53991 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 53992 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 53993 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 53994 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 53995 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0936, + "step": 53996 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1311, + "step": 53997 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0837, + "step": 53998 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 53999 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 54000 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 54001 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 54002 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 54003 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 54004 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 54005 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 54006 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0783, + "step": 54007 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 54008 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 54009 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 54010 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 54011 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0872, + "step": 54012 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.084, + "step": 54013 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 54014 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 54015 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 54016 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 54017 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 54018 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 54019 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0941, + "step": 54020 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0678, + "step": 54021 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 54022 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.089, + "step": 54023 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0895, + "step": 54024 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 54025 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 54026 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 54027 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 54028 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 54029 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 54030 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0941, + "step": 54031 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 54032 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 54033 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0909, + "step": 54034 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 54035 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 54036 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 54037 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0876, + "step": 54038 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 54039 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 54040 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 54041 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 54042 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 54043 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 54044 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 54045 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0796, + "step": 54046 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0875, + "step": 54047 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 54048 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 54049 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0754, + "step": 54050 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0873, + "step": 54051 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0714, + "step": 54052 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 54053 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 54054 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.087, + "step": 54055 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 54056 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 54057 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.084, + "step": 54058 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 54059 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 54060 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 54061 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 54062 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0725, + "step": 54063 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 54064 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0863, + "step": 54065 + }, + { + "epoch": 3.28, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 54066 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 54067 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0864, + "step": 54068 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 54069 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 54070 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 54071 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0943, + "step": 54072 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 54073 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 54074 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 54075 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 54076 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 54077 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0663, + "step": 54078 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 54079 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 54080 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 54081 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0911, + "step": 54082 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 54083 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 54084 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0642, + "step": 54085 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 54086 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 54087 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 54088 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 54089 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 54090 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0864, + "step": 54091 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0759, + "step": 54092 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0817, + "step": 54093 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0853, + "step": 54094 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 54095 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 54096 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.085, + "step": 54097 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 54098 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 54099 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 54100 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 54101 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 54102 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0835, + "step": 54103 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0777, + "step": 54104 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 54105 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 54106 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1469, + "step": 54107 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0759, + "step": 54108 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 54109 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 54110 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 54111 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0839, + "step": 54112 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 54113 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 54114 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 54115 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 54116 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 54117 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 54118 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0866, + "step": 54119 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 54120 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0828, + "step": 54121 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0837, + "step": 54122 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0872, + "step": 54123 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0831, + "step": 54124 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 54125 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 54126 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 54127 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0813, + "step": 54128 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.163, + "step": 54129 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 54130 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 54131 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 54132 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 54133 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 54134 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 54135 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0818, + "step": 54136 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 54137 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 54138 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 54139 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 54140 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 54141 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 54142 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 54143 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0881, + "step": 54144 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1512, + "step": 54145 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 54146 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 54147 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 54148 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 54149 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.078, + "step": 54150 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 54151 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0799, + "step": 54152 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 54153 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 54154 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 54155 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 54156 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 54157 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 54158 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.085, + "step": 54159 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 54160 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 54161 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0807, + "step": 54162 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 54163 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0778, + "step": 54164 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 54165 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 54166 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 54167 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 54168 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0805, + "step": 54169 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 54170 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 54171 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0848, + "step": 54172 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 54173 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 54174 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0895, + "step": 54175 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 54176 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.147, + "step": 54177 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0917, + "step": 54178 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 54179 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 54180 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 54181 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 54182 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 54183 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0829, + "step": 54184 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0779, + "step": 54185 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 54186 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0743, + "step": 54187 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 54188 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.091, + "step": 54189 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 54190 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0859, + "step": 54191 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 54192 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 54193 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 54194 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 54195 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 54196 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1402, + "step": 54197 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0737, + "step": 54198 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0792, + "step": 54199 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 54200 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0827, + "step": 54201 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0878, + "step": 54202 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 54203 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0871, + "step": 54204 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0868, + "step": 54205 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0903, + "step": 54206 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 54207 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 54208 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0918, + "step": 54209 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0916, + "step": 54210 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 54211 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0872, + "step": 54212 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 54213 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 54214 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 54215 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0764, + "step": 54216 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 54217 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 54218 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 54219 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 54220 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 54221 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 54222 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0914, + "step": 54223 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 54224 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 54225 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0674, + "step": 54226 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0825, + "step": 54227 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 54228 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0853, + "step": 54229 + }, + { + "epoch": 3.29, + "learning_rate": 2e-05, + "loss": 0.0915, + "step": 54230 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0794, + "step": 54231 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 54232 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 54233 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 54234 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0871, + "step": 54235 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 54236 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 54237 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 54238 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 54239 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 54240 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0917, + "step": 54241 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 54242 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0804, + "step": 54243 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0789, + "step": 54244 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 54245 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 54246 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 54247 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 54248 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.082, + "step": 54249 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 54250 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 54251 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0835, + "step": 54252 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 54253 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 54254 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 54255 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0812, + "step": 54256 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 54257 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 54258 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 54259 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 54260 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 54261 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 54262 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 54263 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 54264 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 54265 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 54266 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 54267 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 54268 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 54269 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0818, + "step": 54270 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0894, + "step": 54271 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 54272 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0818, + "step": 54273 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 54274 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 54275 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1551, + "step": 54276 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 54277 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 54278 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 54279 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 54280 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 54281 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 54282 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 54283 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0882, + "step": 54284 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 54285 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 54286 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 54287 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 54288 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 54289 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 54290 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 54291 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 54292 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 54293 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 54294 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 54295 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0883, + "step": 54296 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 54297 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 54298 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 54299 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 54300 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 54301 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 54302 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 54303 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 54304 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.08, + "step": 54305 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0857, + "step": 54306 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 54307 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 54308 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 54309 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 54310 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0848, + "step": 54311 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 54312 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 54313 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 54314 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 54315 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0779, + "step": 54316 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 54317 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 54318 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0779, + "step": 54319 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 54320 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 54321 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0797, + "step": 54322 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 54323 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.079, + "step": 54324 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0751, + "step": 54325 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 54326 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 54327 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 54328 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 54329 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.08, + "step": 54330 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 54331 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 54332 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 54333 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 54334 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 54335 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0902, + "step": 54336 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 54337 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0874, + "step": 54338 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 54339 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 54340 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 54341 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 54342 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.127, + "step": 54343 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 54344 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0879, + "step": 54345 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 54346 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 54347 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0803, + "step": 54348 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0848, + "step": 54349 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 54350 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0743, + "step": 54351 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 54352 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 54353 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0898, + "step": 54354 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 54355 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 54356 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 54357 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0915, + "step": 54358 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 54359 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 54360 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 54361 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 54362 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0817, + "step": 54363 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0859, + "step": 54364 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 54365 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 54366 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 54367 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 54368 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 54369 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 54370 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 54371 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 54372 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 54373 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 54374 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 54375 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 54376 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 54377 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0911, + "step": 54378 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0756, + "step": 54379 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0671, + "step": 54380 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 54381 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0816, + "step": 54382 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0834, + "step": 54383 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 54384 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 54385 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 54386 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0872, + "step": 54387 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 54388 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 54389 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 54390 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 54391 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0858, + "step": 54392 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.0878, + "step": 54393 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 54394 + }, + { + "epoch": 3.3, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 54395 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 54396 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0853, + "step": 54397 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1281, + "step": 54398 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 54399 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0865, + "step": 54400 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 54401 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0784, + "step": 54402 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0894, + "step": 54403 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0796, + "step": 54404 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 54405 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0755, + "step": 54406 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 54407 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0794, + "step": 54408 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 54409 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 54410 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 54411 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0857, + "step": 54412 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 54413 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 54414 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 54415 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 54416 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 54417 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 54418 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 54419 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0827, + "step": 54420 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 54421 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 54422 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 54423 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 54424 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 54425 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 54426 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 54427 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 54428 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 54429 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 54430 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 54431 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 54432 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 54433 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 54434 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 54435 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 54436 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0936, + "step": 54437 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.091, + "step": 54438 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 54439 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 54440 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 54441 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 54442 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0879, + "step": 54443 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 54444 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 54445 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 54446 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0903, + "step": 54447 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0804, + "step": 54448 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 54449 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 54450 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 54451 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 54452 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0766, + "step": 54453 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 54454 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0616, + "step": 54455 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 54456 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 54457 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 54458 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 54459 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0915, + "step": 54460 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 54461 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0878, + "step": 54462 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 54463 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0651, + "step": 54464 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 54465 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 54466 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.085, + "step": 54467 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 54468 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 54469 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 54470 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.08, + "step": 54471 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0792, + "step": 54472 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 54473 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 54474 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1304, + "step": 54475 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 54476 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.071, + "step": 54477 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 54478 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 54479 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0876, + "step": 54480 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0809, + "step": 54481 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0902, + "step": 54482 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 54483 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 54484 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 54485 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 54486 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 54487 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 54488 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 54489 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 54490 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.082, + "step": 54491 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 54492 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0906, + "step": 54493 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 54494 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 54495 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 54496 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 54497 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 54498 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0797, + "step": 54499 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0932, + "step": 54500 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0853, + "step": 54501 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 54502 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 54503 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 54504 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 54505 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1398, + "step": 54506 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0875, + "step": 54507 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 54508 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 54509 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0855, + "step": 54510 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 54511 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 54512 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 54513 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1472, + "step": 54514 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 54515 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 54516 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 54517 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 54518 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 54519 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 54520 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0698, + "step": 54521 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 54522 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 54523 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0823, + "step": 54524 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 54525 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 54526 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 54527 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 54528 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 54529 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 54530 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 54531 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 54532 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0815, + "step": 54533 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0773, + "step": 54534 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0826, + "step": 54535 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 54536 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 54537 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 54538 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 54539 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 54540 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 54541 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0826, + "step": 54542 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 54543 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 54544 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 54545 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0873, + "step": 54546 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0771, + "step": 54547 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 54548 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 54549 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 54550 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 54551 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 54552 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 54553 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 54554 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 54555 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0807, + "step": 54556 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0941, + "step": 54557 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 54558 + }, + { + "epoch": 3.31, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 54559 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0763, + "step": 54560 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0867, + "step": 54561 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 54562 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0915, + "step": 54563 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 54564 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 54565 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 54566 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 54567 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 54568 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 54569 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0766, + "step": 54570 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0837, + "step": 54571 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 54572 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 54573 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.089, + "step": 54574 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0823, + "step": 54575 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 54576 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 54577 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 54578 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 54579 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 54580 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 54581 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0936, + "step": 54582 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 54583 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 54584 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0792, + "step": 54585 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 54586 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0872, + "step": 54587 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0742, + "step": 54588 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 54589 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 54590 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 54591 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 54592 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 54593 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 54594 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 54595 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0731, + "step": 54596 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 54597 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0821, + "step": 54598 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0886, + "step": 54599 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0863, + "step": 54600 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 54601 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 54602 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 54603 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0865, + "step": 54604 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0943, + "step": 54605 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 54606 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 54607 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 54608 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 54609 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 54610 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 54611 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 54612 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 54613 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 54614 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0762, + "step": 54615 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 54616 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 54617 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1489, + "step": 54618 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 54619 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0878, + "step": 54620 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 54621 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 54622 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 54623 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 54624 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 54625 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0798, + "step": 54626 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0829, + "step": 54627 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0817, + "step": 54628 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 54629 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0803, + "step": 54630 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0714, + "step": 54631 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 54632 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 54633 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 54634 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0943, + "step": 54635 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 54636 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 54637 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 54638 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0881, + "step": 54639 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0727, + "step": 54640 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0802, + "step": 54641 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 54642 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 54643 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 54644 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 54645 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0838, + "step": 54646 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0887, + "step": 54647 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 54648 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 54649 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 54650 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 54651 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 54652 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0751, + "step": 54653 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0825, + "step": 54654 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 54655 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 54656 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 54657 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 54658 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 54659 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 54660 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 54661 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 54662 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 54663 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0684, + "step": 54664 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 54665 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 54666 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 54667 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 54668 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0847, + "step": 54669 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 54670 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0763, + "step": 54671 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 54672 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0895, + "step": 54673 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 54674 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0851, + "step": 54675 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 54676 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0791, + "step": 54677 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 54678 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 54679 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 54680 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0873, + "step": 54681 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 54682 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.08, + "step": 54683 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 54684 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 54685 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 54686 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 54687 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 54688 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0829, + "step": 54689 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0755, + "step": 54690 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 54691 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 54692 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 54693 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0871, + "step": 54694 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0828, + "step": 54695 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0899, + "step": 54696 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 54697 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 54698 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 54699 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 54700 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 54701 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 54702 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 54703 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0768, + "step": 54704 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 54705 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0776, + "step": 54706 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0779, + "step": 54707 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 54708 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 54709 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 54710 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0837, + "step": 54711 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 54712 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 54713 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0765, + "step": 54714 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 54715 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 54716 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 54717 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0829, + "step": 54718 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.088, + "step": 54719 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 54720 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 54721 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 54722 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0899, + "step": 54723 + }, + { + "epoch": 3.32, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 54724 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 54725 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0816, + "step": 54726 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 54727 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 54728 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 54729 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 54730 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 54731 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0936, + "step": 54732 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0906, + "step": 54733 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 54734 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1546, + "step": 54735 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 54736 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0789, + "step": 54737 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 54738 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 54739 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 54740 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 54741 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1503, + "step": 54742 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0809, + "step": 54743 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 54744 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 54745 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 54746 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 54747 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0911, + "step": 54748 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 54749 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0869, + "step": 54750 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 54751 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 54752 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0876, + "step": 54753 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 54754 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 54755 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0876, + "step": 54756 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 54757 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0704, + "step": 54758 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 54759 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 54760 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 54761 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 54762 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 54763 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 54764 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 54765 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0882, + "step": 54766 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 54767 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0736, + "step": 54768 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0854, + "step": 54769 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 54770 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0748, + "step": 54771 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 54772 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0671, + "step": 54773 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 54774 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0866, + "step": 54775 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0786, + "step": 54776 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 54777 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 54778 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 54779 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0935, + "step": 54780 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0775, + "step": 54781 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0868, + "step": 54782 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 54783 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 54784 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0899, + "step": 54785 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 54786 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 54787 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 54788 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 54789 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0936, + "step": 54790 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 54791 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1381, + "step": 54792 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1475, + "step": 54793 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 54794 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 54795 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0822, + "step": 54796 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 54797 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 54798 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 54799 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0827, + "step": 54800 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 54801 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 54802 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 54803 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0804, + "step": 54804 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 54805 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 54806 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 54807 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 54808 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 54809 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0877, + "step": 54810 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 54811 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 54812 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 54813 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 54814 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 54815 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0836, + "step": 54816 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 54817 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 54818 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 54819 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0721, + "step": 54820 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 54821 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 54822 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 54823 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 54824 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 54825 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 54826 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1411, + "step": 54827 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 54828 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 54829 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1364, + "step": 54830 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 54831 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 54832 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 54833 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0836, + "step": 54834 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0822, + "step": 54835 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 54836 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 54837 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 54838 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 54839 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0714, + "step": 54840 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 54841 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 54842 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 54843 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0871, + "step": 54844 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 54845 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0655, + "step": 54846 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0881, + "step": 54847 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0768, + "step": 54848 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 54849 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 54850 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 54851 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 54852 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 54853 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 54854 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0767, + "step": 54855 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 54856 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 54857 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 54858 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0877, + "step": 54859 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0753, + "step": 54860 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0684, + "step": 54861 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 54862 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1228, + "step": 54863 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 54864 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 54865 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 54866 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 54867 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 54868 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 54869 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.082, + "step": 54870 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 54871 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0738, + "step": 54872 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 54873 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 54874 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 54875 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0825, + "step": 54876 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 54877 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 54878 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0778, + "step": 54879 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 54880 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0853, + "step": 54881 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0844, + "step": 54882 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 54883 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0739, + "step": 54884 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 54885 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 54886 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0787, + "step": 54887 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 54888 + }, + { + "epoch": 3.33, + "learning_rate": 2e-05, + "loss": 0.0841, + "step": 54889 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 54890 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 54891 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0787, + "step": 54892 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 54893 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 54894 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 54895 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 54896 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 54897 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0811, + "step": 54898 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 54899 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 54900 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0848, + "step": 54901 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 54902 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 54903 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 54904 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 54905 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0804, + "step": 54906 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 54907 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 54908 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 54909 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 54910 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0827, + "step": 54911 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 54912 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 54913 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 54914 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 54915 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0819, + "step": 54916 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 54917 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 54918 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1456, + "step": 54919 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0918, + "step": 54920 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0791, + "step": 54921 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 54922 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0936, + "step": 54923 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 54924 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 54925 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 54926 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 54927 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1286, + "step": 54928 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 54929 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 54930 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 54931 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.082, + "step": 54932 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 54933 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 54934 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0891, + "step": 54935 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 54936 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 54937 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0894, + "step": 54938 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 54939 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 54940 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 54941 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 54942 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 54943 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 54944 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0858, + "step": 54945 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0873, + "step": 54946 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 54947 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 54948 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 54949 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 54950 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 54951 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 54952 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0943, + "step": 54953 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0877, + "step": 54954 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 54955 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 54956 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0822, + "step": 54957 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 54958 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 54959 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.08, + "step": 54960 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 54961 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 54962 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 54963 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 54964 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 54965 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 54966 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 54967 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 54968 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 54969 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0853, + "step": 54970 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 54971 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 54972 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0918, + "step": 54973 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 54974 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 54975 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0941, + "step": 54976 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0765, + "step": 54977 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 54978 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 54979 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 54980 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 54981 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 54982 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1495, + "step": 54983 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 54984 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 54985 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0852, + "step": 54986 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.09, + "step": 54987 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 54988 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 54989 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 54990 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 54991 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 54992 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 54993 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 54994 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 54995 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 54996 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 54997 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0812, + "step": 54998 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 54999 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 55000 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 55001 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 55002 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 55003 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0863, + "step": 55004 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0876, + "step": 55005 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0882, + "step": 55006 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 55007 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1543, + "step": 55008 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 55009 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 55010 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 55011 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 55012 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 55013 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 55014 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0831, + "step": 55015 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 55016 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 55017 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 55018 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0803, + "step": 55019 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 55020 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 55021 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 55022 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0761, + "step": 55023 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 55024 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 55025 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.079, + "step": 55026 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 55027 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0936, + "step": 55028 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0802, + "step": 55029 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 55030 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 55031 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 55032 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0935, + "step": 55033 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 55034 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 55035 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 55036 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 55037 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 55038 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0714, + "step": 55039 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 55040 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 55041 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 55042 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.083, + "step": 55043 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 55044 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0894, + "step": 55045 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1435, + "step": 55046 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 55047 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 55048 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0833, + "step": 55049 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 55050 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 55051 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 55052 + }, + { + "epoch": 3.34, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 55053 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0862, + "step": 55054 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0882, + "step": 55055 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0775, + "step": 55056 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 55057 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 55058 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 55059 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 55060 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 55061 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 55062 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 55063 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 55064 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 55065 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 55066 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 55067 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 55068 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 55069 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 55070 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 55071 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 55072 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 55073 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 55074 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 55075 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 55076 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 55077 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 55078 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0845, + "step": 55079 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0791, + "step": 55080 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0836, + "step": 55081 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0876, + "step": 55082 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 55083 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0906, + "step": 55084 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 55085 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 55086 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0709, + "step": 55087 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 55088 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 55089 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 55090 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 55091 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.138, + "step": 55092 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 55093 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 55094 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 55095 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0765, + "step": 55096 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 55097 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0779, + "step": 55098 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 55099 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 55100 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 55101 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 55102 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0695, + "step": 55103 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 55104 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 55105 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 55106 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 55107 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 55108 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0917, + "step": 55109 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 55110 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 55111 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 55112 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 55113 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1366, + "step": 55114 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 55115 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0699, + "step": 55116 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0806, + "step": 55117 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 55118 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 55119 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0906, + "step": 55120 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 55121 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 55122 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 55123 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 55124 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 55125 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0914, + "step": 55126 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.071, + "step": 55127 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 55128 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 55129 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0929, + "step": 55130 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 55131 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 55132 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 55133 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 55134 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0892, + "step": 55135 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0726, + "step": 55136 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0796, + "step": 55137 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0862, + "step": 55138 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 55139 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 55140 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 55141 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 55142 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 55143 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 55144 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 55145 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 55146 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0714, + "step": 55147 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 55148 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 55149 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 55150 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 55151 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0894, + "step": 55152 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 55153 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0777, + "step": 55154 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 55155 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 55156 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 55157 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1377, + "step": 55158 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 55159 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 55160 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0837, + "step": 55161 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 55162 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 55163 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 55164 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 55165 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 55166 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 55167 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 55168 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 55169 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0886, + "step": 55170 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 55171 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 55172 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 55173 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 55174 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 55175 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0863, + "step": 55176 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1348, + "step": 55177 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0856, + "step": 55178 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1465, + "step": 55179 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 55180 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 55181 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 55182 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 55183 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 55184 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 55185 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 55186 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 55187 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0952, + "step": 55188 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 55189 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0929, + "step": 55190 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0877, + "step": 55191 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.091, + "step": 55192 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0863, + "step": 55193 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0895, + "step": 55194 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 55195 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.084, + "step": 55196 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 55197 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0778, + "step": 55198 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 55199 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 55200 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 55201 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 55202 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 55203 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0815, + "step": 55204 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0873, + "step": 55205 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 55206 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 55207 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1293, + "step": 55208 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0671, + "step": 55209 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 55210 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 55211 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 55212 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 55213 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0738, + "step": 55214 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 55215 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 55216 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 55217 + }, + { + "epoch": 3.35, + "learning_rate": 2e-05, + "loss": 0.0943, + "step": 55218 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0932, + "step": 55219 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 55220 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 55221 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 55222 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0872, + "step": 55223 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 55224 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 55225 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 55226 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 55227 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 55228 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0853, + "step": 55229 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0822, + "step": 55230 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 55231 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 55232 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 55233 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 55234 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 55235 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 55236 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 55237 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0932, + "step": 55238 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.088, + "step": 55239 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1447, + "step": 55240 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 55241 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 55242 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 55243 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 55244 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 55245 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 55246 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 55247 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 55248 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0768, + "step": 55249 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 55250 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0901, + "step": 55251 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.08, + "step": 55252 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 55253 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 55254 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 55255 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 55256 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0845, + "step": 55257 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0749, + "step": 55258 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 55259 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 55260 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0735, + "step": 55261 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 55262 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 55263 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 55264 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 55265 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0851, + "step": 55266 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0887, + "step": 55267 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 55268 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 55269 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 55270 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 55271 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0843, + "step": 55272 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 55273 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 55274 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0891, + "step": 55275 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 55276 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 55277 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 55278 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0847, + "step": 55279 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 55280 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 55281 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 55282 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0902, + "step": 55283 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0915, + "step": 55284 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 55285 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 55286 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 55287 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1537, + "step": 55288 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 55289 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 55290 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 55291 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 55292 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 55293 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 55294 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 55295 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 55296 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 55297 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 55298 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 55299 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 55300 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 55301 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0756, + "step": 55302 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 55303 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0864, + "step": 55304 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 55305 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0882, + "step": 55306 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 55307 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 55308 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 55309 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 55310 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 55311 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 55312 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 55313 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0794, + "step": 55314 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0843, + "step": 55315 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0833, + "step": 55316 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 55317 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0826, + "step": 55318 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 55319 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 55320 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 55321 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.073, + "step": 55322 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.091, + "step": 55323 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 55324 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 55325 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 55326 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 55327 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 55328 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0738, + "step": 55329 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 55330 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 55331 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 55332 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0842, + "step": 55333 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 55334 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 55335 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 55336 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0871, + "step": 55337 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0895, + "step": 55338 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0756, + "step": 55339 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 55340 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 55341 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 55342 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 55343 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 55344 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 55345 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 55346 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 55347 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0807, + "step": 55348 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 55349 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 55350 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 55351 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 55352 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 55353 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 55354 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 55355 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 55356 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 55357 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 55358 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 55359 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 55360 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 55361 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.091, + "step": 55362 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 55363 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0851, + "step": 55364 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0915, + "step": 55365 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 55366 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 55367 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 55368 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0902, + "step": 55369 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 55370 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.087, + "step": 55371 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 55372 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 55373 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 55374 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 55375 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 55376 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 55377 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0883, + "step": 55378 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0742, + "step": 55379 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.087, + "step": 55380 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 55381 + }, + { + "epoch": 3.36, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 55382 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 55383 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 55384 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 55385 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 55386 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 55387 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 55388 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0849, + "step": 55389 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 55390 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 55391 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 55392 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0806, + "step": 55393 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0906, + "step": 55394 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 55395 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 55396 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 55397 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 55398 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 55399 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0877, + "step": 55400 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 55401 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0856, + "step": 55402 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0837, + "step": 55403 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0806, + "step": 55404 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0915, + "step": 55405 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 55406 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 55407 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 55408 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 55409 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0817, + "step": 55410 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 55411 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 55412 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 55413 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 55414 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 55415 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 55416 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 55417 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 55418 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0852, + "step": 55419 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 55420 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 55421 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 55422 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 55423 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 55424 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0857, + "step": 55425 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 55426 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 55427 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 55428 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 55429 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0903, + "step": 55430 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 55431 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 55432 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 55433 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1403, + "step": 55434 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0902, + "step": 55435 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 55436 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0801, + "step": 55437 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 55438 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 55439 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0917, + "step": 55440 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 55441 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0755, + "step": 55442 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 55443 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0898, + "step": 55444 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0804, + "step": 55445 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 55446 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 55447 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 55448 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0841, + "step": 55449 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0917, + "step": 55450 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 55451 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0899, + "step": 55452 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 55453 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0852, + "step": 55454 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 55455 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 55456 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 55457 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 55458 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 55459 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0817, + "step": 55460 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 55461 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 55462 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0897, + "step": 55463 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 55464 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0802, + "step": 55465 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0826, + "step": 55466 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 55467 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0818, + "step": 55468 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 55469 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 55470 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 55471 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 55472 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 55473 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0866, + "step": 55474 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 55475 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0879, + "step": 55476 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 55477 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 55478 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 55479 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 55480 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 55481 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0751, + "step": 55482 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 55483 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 55484 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 55485 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0762, + "step": 55486 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 55487 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 55488 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0783, + "step": 55489 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 55490 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 55491 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0859, + "step": 55492 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 55493 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0782, + "step": 55494 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 55495 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 55496 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 55497 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0862, + "step": 55498 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 55499 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 55500 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 55501 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 55502 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0662, + "step": 55503 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0888, + "step": 55504 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0877, + "step": 55505 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0847, + "step": 55506 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.081, + "step": 55507 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 55508 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 55509 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 55510 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0853, + "step": 55511 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 55512 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 55513 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 55514 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 55515 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 55516 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 55517 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0855, + "step": 55518 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.074, + "step": 55519 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 55520 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 55521 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0793, + "step": 55522 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0754, + "step": 55523 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0849, + "step": 55524 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 55525 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 55526 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 55527 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 55528 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 55529 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 55530 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 55531 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 55532 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 55533 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 55534 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 55535 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 55536 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 55537 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 55538 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 55539 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 55540 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 55541 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 55542 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 55543 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 55544 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 55545 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.0834, + "step": 55546 + }, + { + "epoch": 3.37, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 55547 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 55548 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 55549 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0859, + "step": 55550 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 55551 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0754, + "step": 55552 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 55553 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 55554 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 55555 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 55556 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 55557 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 55558 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0877, + "step": 55559 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0856, + "step": 55560 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 55561 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 55562 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 55563 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 55564 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 55565 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 55566 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 55567 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 55568 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 55569 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 55570 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 55571 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 55572 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 55573 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 55574 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0857, + "step": 55575 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1415, + "step": 55576 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 55577 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 55578 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0843, + "step": 55579 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 55580 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 55581 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0738, + "step": 55582 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 55583 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.085, + "step": 55584 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 55585 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 55586 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 55587 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0832, + "step": 55588 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 55589 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 55590 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 55591 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 55592 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 55593 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0916, + "step": 55594 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0857, + "step": 55595 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 55596 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 55597 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 55598 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 55599 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0871, + "step": 55600 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0916, + "step": 55601 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 55602 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 55603 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 55604 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 55605 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 55606 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0909, + "step": 55607 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0882, + "step": 55608 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 55609 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0915, + "step": 55610 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 55611 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0775, + "step": 55612 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 55613 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0894, + "step": 55614 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 55615 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 55616 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0894, + "step": 55617 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0882, + "step": 55618 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0838, + "step": 55619 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 55620 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0777, + "step": 55621 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 55622 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0952, + "step": 55623 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 55624 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.088, + "step": 55625 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 55626 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 55627 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0761, + "step": 55628 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 55629 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 55630 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 55631 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 55632 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0802, + "step": 55633 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 55634 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 55635 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 55636 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0852, + "step": 55637 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 55638 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0851, + "step": 55639 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 55640 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 55641 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0918, + "step": 55642 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0784, + "step": 55643 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 55644 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0917, + "step": 55645 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 55646 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 55647 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 55648 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 55649 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0753, + "step": 55650 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0817, + "step": 55651 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 55652 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 55653 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 55654 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 55655 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 55656 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 55657 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.084, + "step": 55658 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 55659 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 55660 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 55661 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 55662 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0834, + "step": 55663 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 55664 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0858, + "step": 55665 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 55666 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 55667 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 55668 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 55669 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 55670 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 55671 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 55672 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 55673 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 55674 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0826, + "step": 55675 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 55676 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 55677 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0871, + "step": 55678 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 55679 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0952, + "step": 55680 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0943, + "step": 55681 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0898, + "step": 55682 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 55683 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 55684 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0899, + "step": 55685 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 55686 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 55687 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0914, + "step": 55688 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.082, + "step": 55689 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 55690 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 55691 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 55692 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 55693 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 55694 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 55695 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 55696 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 55697 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 55698 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 55699 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 55700 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0845, + "step": 55701 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 55702 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 55703 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 55704 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 55705 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0909, + "step": 55706 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 55707 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 55708 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 55709 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 55710 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 55711 + }, + { + "epoch": 3.38, + "learning_rate": 2e-05, + "loss": 0.0865, + "step": 55712 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 55713 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 55714 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1385, + "step": 55715 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 55716 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0877, + "step": 55717 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0863, + "step": 55718 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 55719 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0799, + "step": 55720 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 55721 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1607, + "step": 55722 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 55723 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 55724 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0889, + "step": 55725 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 55726 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 55727 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0878, + "step": 55728 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 55729 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0871, + "step": 55730 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 55731 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 55732 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 55733 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0687, + "step": 55734 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0887, + "step": 55735 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 55736 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 55737 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 55738 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0834, + "step": 55739 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0803, + "step": 55740 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 55741 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0806, + "step": 55742 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0802, + "step": 55743 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 55744 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 55745 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 55746 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 55747 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0853, + "step": 55748 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 55749 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 55750 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 55751 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 55752 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 55753 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0837, + "step": 55754 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0858, + "step": 55755 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 55756 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1469, + "step": 55757 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 55758 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 55759 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 55760 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0883, + "step": 55761 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0821, + "step": 55762 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 55763 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 55764 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 55765 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 55766 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0865, + "step": 55767 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 55768 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 55769 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 55770 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0856, + "step": 55771 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 55772 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0806, + "step": 55773 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0849, + "step": 55774 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.078, + "step": 55775 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 55776 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 55777 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 55778 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0847, + "step": 55779 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 55780 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0721, + "step": 55781 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 55782 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1476, + "step": 55783 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 55784 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 55785 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 55786 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 55787 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 55788 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 55789 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 55790 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 55791 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 55792 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0906, + "step": 55793 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 55794 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.079, + "step": 55795 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 55796 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 55797 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 55798 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0777, + "step": 55799 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 55800 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.088, + "step": 55801 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 55802 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0858, + "step": 55803 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 55804 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 55805 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 55806 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 55807 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 55808 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 55809 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 55810 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 55811 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 55812 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 55813 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 55814 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 55815 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 55816 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 55817 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0906, + "step": 55818 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 55819 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0864, + "step": 55820 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 55821 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 55822 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 55823 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0903, + "step": 55824 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 55825 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 55826 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 55827 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0779, + "step": 55828 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 55829 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 55830 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0822, + "step": 55831 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 55832 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 55833 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 55834 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 55835 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.087, + "step": 55836 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 55837 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 55838 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0693, + "step": 55839 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 55840 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 55841 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 55842 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 55843 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 55844 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 55845 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 55846 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0796, + "step": 55847 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1491, + "step": 55848 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 55849 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 55850 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 55851 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 55852 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 55853 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0687, + "step": 55854 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.073, + "step": 55855 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 55856 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0891, + "step": 55857 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0895, + "step": 55858 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0825, + "step": 55859 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 55860 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 55861 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 55862 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0936, + "step": 55863 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 55864 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 55865 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0952, + "step": 55866 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0683, + "step": 55867 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 55868 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 55869 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 55870 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.078, + "step": 55871 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0843, + "step": 55872 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 55873 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.0783, + "step": 55874 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 55875 + }, + { + "epoch": 3.39, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 55876 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 55877 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0891, + "step": 55878 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 55879 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 55880 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0911, + "step": 55881 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 55882 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 55883 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 55884 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0916, + "step": 55885 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 55886 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0713, + "step": 55887 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 55888 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 55889 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 55890 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 55891 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 55892 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0837, + "step": 55893 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 55894 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 55895 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0755, + "step": 55896 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 55897 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0862, + "step": 55898 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0875, + "step": 55899 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 55900 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 55901 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 55902 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 55903 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 55904 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 55905 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 55906 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 55907 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 55908 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 55909 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 55910 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 55911 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 55912 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 55913 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 55914 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 55915 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 55916 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 55917 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 55918 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 55919 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 55920 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 55921 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 55922 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0889, + "step": 55923 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 55924 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 55925 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0877, + "step": 55926 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 55927 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 55928 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 55929 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1456, + "step": 55930 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0883, + "step": 55931 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 55932 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 55933 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 55934 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 55935 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 55936 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 55937 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0786, + "step": 55938 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0902, + "step": 55939 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 55940 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 55941 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0952, + "step": 55942 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 55943 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 55944 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 55945 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 55946 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 55947 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0941, + "step": 55948 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 55949 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 55950 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 55951 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 55952 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 55953 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 55954 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 55955 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 55956 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 55957 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 55958 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 55959 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 55960 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0851, + "step": 55961 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0744, + "step": 55962 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 55963 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0899, + "step": 55964 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 55965 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 55966 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 55967 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 55968 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 55969 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 55970 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0875, + "step": 55971 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 55972 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 55973 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 55974 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 55975 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 55976 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 55977 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 55978 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0866, + "step": 55979 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0744, + "step": 55980 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.089, + "step": 55981 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0781, + "step": 55982 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 55983 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0852, + "step": 55984 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 55985 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 55986 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 55987 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 55988 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 55989 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 55990 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 55991 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0886, + "step": 55992 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0792, + "step": 55993 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 55994 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 55995 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0772, + "step": 55996 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 55997 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0772, + "step": 55998 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 55999 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 56000 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0714, + "step": 56001 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 56002 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0821, + "step": 56003 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 56004 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 56005 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 56006 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0763, + "step": 56007 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 56008 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 56009 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 56010 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 56011 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 56012 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 56013 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0891, + "step": 56014 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 56015 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 56016 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0871, + "step": 56017 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 56018 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0812, + "step": 56019 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.072, + "step": 56020 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 56021 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 56022 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 56023 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 56024 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 56025 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 56026 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 56027 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 56028 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 56029 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 56030 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 56031 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 56032 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 56033 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 56034 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 56035 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 56036 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 56037 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 56038 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 56039 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 56040 + }, + { + "epoch": 3.4, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 56041 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 56042 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 56043 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0821, + "step": 56044 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 56045 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 56046 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 56047 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 56048 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 56049 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 56050 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 56051 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0915, + "step": 56052 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 56053 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 56054 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 56055 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0869, + "step": 56056 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0909, + "step": 56057 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 56058 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 56059 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 56060 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0811, + "step": 56061 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0902, + "step": 56062 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0707, + "step": 56063 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0911, + "step": 56064 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 56065 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0817, + "step": 56066 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 56067 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 56068 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0932, + "step": 56069 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 56070 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0808, + "step": 56071 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0918, + "step": 56072 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 56073 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.082, + "step": 56074 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1474, + "step": 56075 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 56076 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 56077 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0857, + "step": 56078 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 56079 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 56080 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0814, + "step": 56081 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 56082 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 56083 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 56084 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 56085 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 56086 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 56087 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 56088 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 56089 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 56090 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 56091 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 56092 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 56093 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0816, + "step": 56094 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 56095 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 56096 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0829, + "step": 56097 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 56098 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 56099 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 56100 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0875, + "step": 56101 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 56102 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 56103 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 56104 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 56105 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 56106 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 56107 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0796, + "step": 56108 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 56109 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0773, + "step": 56110 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 56111 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0863, + "step": 56112 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 56113 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 56114 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 56115 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 56116 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 56117 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0929, + "step": 56118 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 56119 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0716, + "step": 56120 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 56121 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 56122 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0872, + "step": 56123 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 56124 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 56125 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 56126 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1291, + "step": 56127 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 56128 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0846, + "step": 56129 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 56130 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 56131 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 56132 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 56133 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0764, + "step": 56134 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 56135 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 56136 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0897, + "step": 56137 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 56138 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0878, + "step": 56139 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 56140 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 56141 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 56142 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0909, + "step": 56143 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 56144 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 56145 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 56146 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0816, + "step": 56147 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 56148 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0887, + "step": 56149 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 56150 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 56151 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 56152 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 56153 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 56154 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 56155 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 56156 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 56157 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 56158 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 56159 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0811, + "step": 56160 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 56161 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0737, + "step": 56162 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0863, + "step": 56163 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 56164 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 56165 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 56166 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 56167 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0876, + "step": 56168 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 56169 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0877, + "step": 56170 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 56171 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0803, + "step": 56172 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0744, + "step": 56173 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 56174 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 56175 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 56176 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 56177 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 56178 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0758, + "step": 56179 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 56180 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 56181 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 56182 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0839, + "step": 56183 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0929, + "step": 56184 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 56185 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 56186 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 56187 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 56188 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 56189 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 56190 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 56191 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0845, + "step": 56192 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 56193 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 56194 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 56195 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 56196 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 56197 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 56198 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 56199 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 56200 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 56201 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 56202 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.09, + "step": 56203 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 56204 + }, + { + "epoch": 3.41, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 56205 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 56206 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 56207 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 56208 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 56209 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0882, + "step": 56210 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 56211 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0874, + "step": 56212 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 56213 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 56214 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 56215 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0859, + "step": 56216 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 56217 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0794, + "step": 56218 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 56219 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 56220 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 56221 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 56222 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 56223 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0798, + "step": 56224 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0805, + "step": 56225 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0858, + "step": 56226 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0868, + "step": 56227 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0909, + "step": 56228 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0803, + "step": 56229 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 56230 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 56231 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0794, + "step": 56232 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 56233 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 56234 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 56235 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 56236 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.087, + "step": 56237 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 56238 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 56239 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1372, + "step": 56240 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 56241 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 56242 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 56243 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0794, + "step": 56244 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 56245 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 56246 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 56247 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 56248 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 56249 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 56250 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 56251 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 56252 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.084, + "step": 56253 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 56254 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 56255 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 56256 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 56257 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 56258 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0898, + "step": 56259 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0839, + "step": 56260 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 56261 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 56262 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 56263 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0773, + "step": 56264 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 56265 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.091, + "step": 56266 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 56267 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 56268 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 56269 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 56270 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 56271 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 56272 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0833, + "step": 56273 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0909, + "step": 56274 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0916, + "step": 56275 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 56276 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 56277 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0849, + "step": 56278 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 56279 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 56280 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 56281 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 56282 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 56283 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 56284 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 56285 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 56286 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 56287 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0691, + "step": 56288 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 56289 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1231, + "step": 56290 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 56291 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 56292 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 56293 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 56294 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0771, + "step": 56295 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 56296 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 56297 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 56298 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 56299 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 56300 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 56301 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0892, + "step": 56302 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 56303 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0646, + "step": 56304 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0738, + "step": 56305 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0967, + "step": 56306 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0867, + "step": 56307 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 56308 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 56309 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 56310 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 56311 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0807, + "step": 56312 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 56313 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 56314 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 56315 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 56316 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0917, + "step": 56317 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.083, + "step": 56318 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 56319 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 56320 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 56321 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 56322 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0811, + "step": 56323 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 56324 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.089, + "step": 56325 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 56326 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0854, + "step": 56327 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 56328 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 56329 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 56330 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 56331 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 56332 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 56333 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 56334 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 56335 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 56336 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 56337 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 56338 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 56339 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 56340 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 56341 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 56342 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 56343 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 56344 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.09, + "step": 56345 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0857, + "step": 56346 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 56347 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 56348 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 56349 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 56350 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 56351 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 56352 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 56353 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 56354 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0886, + "step": 56355 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 56356 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 56357 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 56358 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 56359 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 56360 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 56361 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 56362 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 56363 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0935, + "step": 56364 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 56365 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 56366 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.0818, + "step": 56367 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 56368 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 56369 + }, + { + "epoch": 3.42, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 56370 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 56371 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 56372 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 56373 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 56374 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 56375 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 56376 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 56377 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 56378 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1376, + "step": 56379 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 56380 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 56381 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0831, + "step": 56382 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 56383 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 56384 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 56385 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 56386 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0832, + "step": 56387 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 56388 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 56389 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0782, + "step": 56390 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0826, + "step": 56391 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 56392 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 56393 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 56394 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 56395 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0943, + "step": 56396 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 56397 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 56398 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 56399 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 56400 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 56401 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 56402 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 56403 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 56404 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 56405 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 56406 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0852, + "step": 56407 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 56408 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0871, + "step": 56409 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 56410 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0883, + "step": 56411 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 56412 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 56413 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0898, + "step": 56414 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 56415 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 56416 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 56417 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 56418 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 56419 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 56420 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 56421 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0728, + "step": 56422 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 56423 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0848, + "step": 56424 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 56425 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.084, + "step": 56426 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0814, + "step": 56427 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 56428 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 56429 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 56430 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 56431 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0878, + "step": 56432 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1159, + "step": 56433 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0675, + "step": 56434 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 56435 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 56436 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0819, + "step": 56437 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0891, + "step": 56438 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 56439 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 56440 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 56441 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 56442 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 56443 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 56444 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 56445 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 56446 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 56447 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 56448 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 56449 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 56450 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 56451 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 56452 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 56453 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 56454 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 56455 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.07, + "step": 56456 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 56457 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 56458 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1428, + "step": 56459 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0851, + "step": 56460 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 56461 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0655, + "step": 56462 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 56463 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 56464 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 56465 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 56466 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0868, + "step": 56467 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 56468 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 56469 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0935, + "step": 56470 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 56471 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 56472 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 56473 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 56474 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 56475 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 56476 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 56477 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 56478 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0894, + "step": 56479 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 56480 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 56481 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 56482 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0869, + "step": 56483 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 56484 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 56485 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 56486 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 56487 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 56488 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 56489 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 56490 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 56491 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0807, + "step": 56492 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 56493 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 56494 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 56495 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0826, + "step": 56496 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0967, + "step": 56497 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 56498 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0743, + "step": 56499 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 56500 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 56501 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 56502 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 56503 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0822, + "step": 56504 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0848, + "step": 56505 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 56506 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0732, + "step": 56507 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 56508 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 56509 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 56510 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 56511 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 56512 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 56513 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 56514 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 56515 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 56516 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 56517 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 56518 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 56519 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 56520 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 56521 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 56522 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 56523 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 56524 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 56525 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 56526 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 56527 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 56528 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 56529 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 56530 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0789, + "step": 56531 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.0886, + "step": 56532 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 56533 + }, + { + "epoch": 3.43, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 56534 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 56535 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 56536 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0827, + "step": 56537 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 56538 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 56539 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 56540 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0741, + "step": 56541 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 56542 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 56543 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 56544 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 56545 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 56546 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 56547 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 56548 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 56549 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 56550 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 56551 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 56552 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 56553 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0967, + "step": 56554 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 56555 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0819, + "step": 56556 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 56557 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 56558 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0799, + "step": 56559 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0915, + "step": 56560 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0881, + "step": 56561 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 56562 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 56563 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 56564 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 56565 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 56566 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 56567 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0758, + "step": 56568 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0819, + "step": 56569 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 56570 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 56571 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 56572 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 56573 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 56574 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0852, + "step": 56575 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 56576 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 56577 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 56578 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 56579 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 56580 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 56581 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 56582 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 56583 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 56584 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 56585 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0802, + "step": 56586 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 56587 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 56588 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0808, + "step": 56589 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0881, + "step": 56590 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0839, + "step": 56591 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 56592 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 56593 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 56594 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 56595 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 56596 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 56597 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0916, + "step": 56598 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 56599 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 56600 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 56601 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 56602 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 56603 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 56604 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.091, + "step": 56605 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 56606 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 56607 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 56608 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 56609 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0789, + "step": 56610 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 56611 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 56612 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 56613 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0941, + "step": 56614 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0892, + "step": 56615 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 56616 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.091, + "step": 56617 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0876, + "step": 56618 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 56619 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 56620 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 56621 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 56622 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 56623 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0887, + "step": 56624 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 56625 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 56626 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 56627 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 56628 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 56629 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 56630 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0895, + "step": 56631 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0827, + "step": 56632 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 56633 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0881, + "step": 56634 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 56635 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 56636 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 56637 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 56638 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 56639 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 56640 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0786, + "step": 56641 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 56642 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 56643 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 56644 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1509, + "step": 56645 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 56646 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 56647 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 56648 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 56649 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0859, + "step": 56650 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 56651 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 56652 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 56653 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 56654 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 56655 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 56656 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 56657 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 56658 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 56659 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 56660 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 56661 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 56662 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0819, + "step": 56663 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 56664 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 56665 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 56666 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 56667 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 56668 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0796, + "step": 56669 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 56670 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 56671 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 56672 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 56673 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0783, + "step": 56674 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0604, + "step": 56675 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0837, + "step": 56676 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 56677 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 56678 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0848, + "step": 56679 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0849, + "step": 56680 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0823, + "step": 56681 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 56682 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 56683 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0806, + "step": 56684 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 56685 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 56686 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 56687 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0882, + "step": 56688 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0658, + "step": 56689 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0795, + "step": 56690 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 56691 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0853, + "step": 56692 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 56693 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 56694 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 56695 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 56696 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 56697 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 56698 + }, + { + "epoch": 3.44, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 56699 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 56700 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 56701 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 56702 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 56703 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 56704 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 56705 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 56706 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 56707 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 56708 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 56709 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0797, + "step": 56710 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0715, + "step": 56711 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 56712 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 56713 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 56714 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0671, + "step": 56715 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1566, + "step": 56716 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0895, + "step": 56717 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 56718 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 56719 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 56720 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 56721 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 56722 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 56723 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 56724 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 56725 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0842, + "step": 56726 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 56727 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 56728 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0672, + "step": 56729 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0887, + "step": 56730 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 56731 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 56732 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 56733 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 56734 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 56735 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0763, + "step": 56736 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 56737 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 56738 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 56739 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 56740 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 56741 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 56742 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0754, + "step": 56743 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 56744 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 56745 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 56746 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 56747 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 56748 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.081, + "step": 56749 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 56750 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 56751 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 56752 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 56753 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.09, + "step": 56754 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 56755 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 56756 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 56757 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 56758 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 56759 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 56760 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 56761 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 56762 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 56763 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 56764 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 56765 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0807, + "step": 56766 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 56767 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 56768 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 56769 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 56770 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 56771 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 56772 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0699, + "step": 56773 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 56774 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0832, + "step": 56775 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 56776 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 56777 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 56778 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 56779 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 56780 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 56781 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 56782 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 56783 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 56784 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 56785 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0865, + "step": 56786 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 56787 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0841, + "step": 56788 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 56789 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1341, + "step": 56790 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0898, + "step": 56791 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 56792 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0804, + "step": 56793 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 56794 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 56795 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 56796 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 56797 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 56798 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 56799 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 56800 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 56801 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 56802 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 56803 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 56804 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 56805 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 56806 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0952, + "step": 56807 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 56808 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1253, + "step": 56809 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0929, + "step": 56810 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 56811 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0812, + "step": 56812 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 56813 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0783, + "step": 56814 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 56815 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 56816 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 56817 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 56818 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0879, + "step": 56819 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.088, + "step": 56820 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0834, + "step": 56821 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 56822 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.131, + "step": 56823 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 56824 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 56825 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 56826 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 56827 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 56828 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0849, + "step": 56829 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0887, + "step": 56830 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0849, + "step": 56831 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 56832 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 56833 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 56834 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1393, + "step": 56835 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 56836 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 56837 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 56838 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 56839 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0734, + "step": 56840 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 56841 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0755, + "step": 56842 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 56843 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 56844 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 56845 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 56846 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 56847 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0791, + "step": 56848 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 56849 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 56850 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0804, + "step": 56851 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 56852 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 56853 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 56854 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0862, + "step": 56855 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 56856 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 56857 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 56858 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0889, + "step": 56859 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 56860 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 56861 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 56862 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 56863 + }, + { + "epoch": 3.45, + "learning_rate": 2e-05, + "loss": 0.0798, + "step": 56864 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.161, + "step": 56865 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.089, + "step": 56866 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 56867 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 56868 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0717, + "step": 56869 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 56870 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 56871 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0856, + "step": 56872 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 56873 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 56874 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 56875 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 56876 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 56877 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 56878 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 56879 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 56880 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 56881 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.087, + "step": 56882 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 56883 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0742, + "step": 56884 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0899, + "step": 56885 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 56886 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 56887 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 56888 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 56889 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 56890 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 56891 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 56892 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 56893 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 56894 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 56895 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 56896 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0766, + "step": 56897 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 56898 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0906, + "step": 56899 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 56900 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 56901 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 56902 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0831, + "step": 56903 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1422, + "step": 56904 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 56905 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 56906 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0841, + "step": 56907 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 56908 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 56909 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 56910 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1327, + "step": 56911 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0872, + "step": 56912 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0867, + "step": 56913 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 56914 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 56915 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0892, + "step": 56916 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 56917 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 56918 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 56919 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0902, + "step": 56920 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0842, + "step": 56921 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 56922 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0839, + "step": 56923 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 56924 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 56925 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 56926 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 56927 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 56928 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 56929 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 56930 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 56931 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 56932 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 56933 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 56934 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 56935 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 56936 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 56937 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 56938 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 56939 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 56940 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0871, + "step": 56941 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 56942 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0909, + "step": 56943 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 56944 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0726, + "step": 56945 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 56946 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 56947 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 56948 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 56949 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0936, + "step": 56950 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 56951 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0828, + "step": 56952 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 56953 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1419, + "step": 56954 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 56955 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 56956 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 56957 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0765, + "step": 56958 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 56959 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0935, + "step": 56960 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 56961 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 56962 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0901, + "step": 56963 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 56964 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 56965 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 56966 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 56967 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 56968 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 56969 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 56970 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 56971 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 56972 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 56973 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 56974 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 56975 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 56976 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0887, + "step": 56977 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0876, + "step": 56978 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0952, + "step": 56979 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 56980 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 56981 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0867, + "step": 56982 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0719, + "step": 56983 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 56984 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 56985 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 56986 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 56987 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 56988 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 56989 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 56990 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 56991 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 56992 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 56993 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 56994 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0967, + "step": 56995 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 56996 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 56997 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0829, + "step": 56998 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 56999 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 57000 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0894, + "step": 57001 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 57002 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 57003 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 57004 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 57005 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 57006 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0666, + "step": 57007 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 57008 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 57009 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 57010 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 57011 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 57012 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0881, + "step": 57013 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 57014 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.083, + "step": 57015 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0857, + "step": 57016 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0855, + "step": 57017 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 57018 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 57019 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0766, + "step": 57020 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 57021 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 57022 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 57023 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0872, + "step": 57024 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 57025 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 57026 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 57027 + }, + { + "epoch": 3.46, + "learning_rate": 2e-05, + "loss": 0.0895, + "step": 57028 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0813, + "step": 57029 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 57030 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 57031 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 57032 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 57033 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 57034 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 57035 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 57036 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0902, + "step": 57037 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 57038 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 57039 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 57040 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 57041 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 57042 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 57043 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 57044 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 57045 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 57046 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 57047 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 57048 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 57049 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 57050 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 57051 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 57052 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.087, + "step": 57053 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0898, + "step": 57054 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 57055 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 57056 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 57057 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 57058 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0881, + "step": 57059 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0891, + "step": 57060 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 57061 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 57062 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 57063 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0633, + "step": 57064 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 57065 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 57066 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1424, + "step": 57067 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 57068 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 57069 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 57070 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 57071 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 57072 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0936, + "step": 57073 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 57074 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 57075 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 57076 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 57077 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0935, + "step": 57078 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0777, + "step": 57079 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 57080 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 57081 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0868, + "step": 57082 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 57083 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 57084 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 57085 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 57086 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 57087 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 57088 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 57089 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 57090 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 57091 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 57092 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 57093 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 57094 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0776, + "step": 57095 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 57096 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 57097 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 57098 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0647, + "step": 57099 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 57100 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 57101 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0914, + "step": 57102 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0871, + "step": 57103 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 57104 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 57105 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 57106 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 57107 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1297, + "step": 57108 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.135, + "step": 57109 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 57110 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 57111 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 57112 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 57113 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 57114 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 57115 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 57116 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 57117 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 57118 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 57119 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 57120 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0871, + "step": 57121 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 57122 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0841, + "step": 57123 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 57124 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 57125 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 57126 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0819, + "step": 57127 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 57128 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 57129 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0856, + "step": 57130 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 57131 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 57132 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0882, + "step": 57133 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 57134 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0882, + "step": 57135 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 57136 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 57137 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 57138 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 57139 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 57140 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 57141 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 57142 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 57143 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1201, + "step": 57144 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 57145 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 57146 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0841, + "step": 57147 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 57148 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 57149 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 57150 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 57151 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0916, + "step": 57152 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 57153 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 57154 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1448, + "step": 57155 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 57156 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 57157 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 57158 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0854, + "step": 57159 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0909, + "step": 57160 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 57161 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 57162 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0823, + "step": 57163 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 57164 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0838, + "step": 57165 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.088, + "step": 57166 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 57167 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 57168 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 57169 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.083, + "step": 57170 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 57171 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 57172 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0892, + "step": 57173 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 57174 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 57175 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0819, + "step": 57176 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0943, + "step": 57177 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 57178 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 57179 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 57180 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 57181 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0827, + "step": 57182 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 57183 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 57184 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 57185 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 57186 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.0803, + "step": 57187 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.088, + "step": 57188 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 57189 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 57190 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 57191 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 57192 + }, + { + "epoch": 3.47, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 57193 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0862, + "step": 57194 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 57195 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 57196 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 57197 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 57198 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 57199 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1229, + "step": 57200 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1433, + "step": 57201 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0883, + "step": 57202 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 57203 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 57204 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0735, + "step": 57205 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0789, + "step": 57206 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0781, + "step": 57207 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 57208 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 57209 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0771, + "step": 57210 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 57211 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 57212 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 57213 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 57214 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 57215 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 57216 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0742, + "step": 57217 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 57218 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0796, + "step": 57219 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 57220 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 57221 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0871, + "step": 57222 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 57223 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 57224 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 57225 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 57226 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0732, + "step": 57227 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 57228 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0872, + "step": 57229 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 57230 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 57231 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0902, + "step": 57232 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1604, + "step": 57233 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0769, + "step": 57234 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0811, + "step": 57235 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0875, + "step": 57236 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0943, + "step": 57237 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 57238 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0674, + "step": 57239 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0886, + "step": 57240 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 57241 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 57242 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 57243 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 57244 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 57245 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 57246 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 57247 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0895, + "step": 57248 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 57249 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 57250 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 57251 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 57252 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 57253 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0664, + "step": 57254 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 57255 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 57256 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 57257 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0784, + "step": 57258 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0941, + "step": 57259 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 57260 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 57261 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0756, + "step": 57262 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 57263 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 57264 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 57265 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 57266 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 57267 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 57268 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0936, + "step": 57269 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 57270 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0808, + "step": 57271 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 57272 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 57273 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0814, + "step": 57274 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 57275 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 57276 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0784, + "step": 57277 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 57278 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 57279 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 57280 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 57281 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 57282 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 57283 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0806, + "step": 57284 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 57285 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 57286 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0862, + "step": 57287 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 57288 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 57289 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 57290 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 57291 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 57292 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 57293 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0777, + "step": 57294 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 57295 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 57296 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 57297 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0929, + "step": 57298 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 57299 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 57300 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 57301 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 57302 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0857, + "step": 57303 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0821, + "step": 57304 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1219, + "step": 57305 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 57306 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 57307 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 57308 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 57309 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 57310 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.091, + "step": 57311 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 57312 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0943, + "step": 57313 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0889, + "step": 57314 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 57315 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 57316 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0849, + "step": 57317 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 57318 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0863, + "step": 57319 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 57320 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 57321 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 57322 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0843, + "step": 57323 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0802, + "step": 57324 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 57325 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0829, + "step": 57326 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1365, + "step": 57327 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 57328 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 57329 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 57330 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 57331 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 57332 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 57333 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 57334 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 57335 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 57336 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 57337 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 57338 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 57339 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 57340 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0844, + "step": 57341 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0814, + "step": 57342 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 57343 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 57344 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 57345 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 57346 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 57347 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 57348 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 57349 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 57350 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 57351 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 57352 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 57353 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 57354 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 57355 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 57356 + }, + { + "epoch": 3.48, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 57357 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 57358 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 57359 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.085, + "step": 57360 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 57361 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 57362 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 57363 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 57364 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0817, + "step": 57365 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 57366 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0836, + "step": 57367 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 57368 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 57369 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 57370 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 57371 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 57372 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 57373 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 57374 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 57375 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 57376 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.074, + "step": 57377 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 57378 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 57379 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0891, + "step": 57380 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 57381 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 57382 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 57383 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0831, + "step": 57384 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 57385 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0878, + "step": 57386 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 57387 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0778, + "step": 57388 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 57389 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 57390 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 57391 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 57392 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 57393 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 57394 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1278, + "step": 57395 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0903, + "step": 57396 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0865, + "step": 57397 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0894, + "step": 57398 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 57399 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0901, + "step": 57400 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 57401 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0855, + "step": 57402 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 57403 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 57404 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 57405 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.089, + "step": 57406 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 57407 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 57408 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 57409 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0914, + "step": 57410 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 57411 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 57412 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 57413 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 57414 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 57415 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 57416 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 57417 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 57418 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 57419 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 57420 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0867, + "step": 57421 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 57422 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 57423 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 57424 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 57425 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1382, + "step": 57426 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 57427 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 57428 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 57429 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 57430 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 57431 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 57432 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 57433 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 57434 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 57435 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 57436 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 57437 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 57438 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 57439 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0881, + "step": 57440 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 57441 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0894, + "step": 57442 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.085, + "step": 57443 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 57444 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 57445 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 57446 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0732, + "step": 57447 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 57448 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0914, + "step": 57449 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 57450 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 57451 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0868, + "step": 57452 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 57453 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 57454 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0837, + "step": 57455 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 57456 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 57457 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 57458 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 57459 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 57460 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0769, + "step": 57461 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0842, + "step": 57462 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0842, + "step": 57463 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 57464 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 57465 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 57466 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 57467 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 57468 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 57469 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0848, + "step": 57470 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.082, + "step": 57471 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0943, + "step": 57472 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 57473 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0871, + "step": 57474 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 57475 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0888, + "step": 57476 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0876, + "step": 57477 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 57478 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 57479 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 57480 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 57481 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.077, + "step": 57482 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 57483 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 57484 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0837, + "step": 57485 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 57486 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 57487 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 57488 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0813, + "step": 57489 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 57490 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0839, + "step": 57491 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 57492 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 57493 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 57494 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 57495 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 57496 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0851, + "step": 57497 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 57498 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0854, + "step": 57499 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 57500 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 57501 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 57502 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0892, + "step": 57503 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0813, + "step": 57504 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0667, + "step": 57505 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 57506 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 57507 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0889, + "step": 57508 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 57509 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1225, + "step": 57510 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 57511 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 57512 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0863, + "step": 57513 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 57514 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 57515 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 57516 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 57517 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 57518 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 57519 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 57520 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 57521 + }, + { + "epoch": 3.49, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 57522 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 57523 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 57524 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 57525 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 57526 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0842, + "step": 57527 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 57528 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 57529 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 57530 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0798, + "step": 57531 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0941, + "step": 57532 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 57533 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 57534 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0929, + "step": 57535 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 57536 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 57537 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 57538 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 57539 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0676, + "step": 57540 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 57541 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0876, + "step": 57542 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 57543 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 57544 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0776, + "step": 57545 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0943, + "step": 57546 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 57547 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 57548 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 57549 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0903, + "step": 57550 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 57551 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 57552 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 57553 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1258, + "step": 57554 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 57555 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 57556 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 57557 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 57558 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 57559 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 57560 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 57561 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 57562 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 57563 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0888, + "step": 57564 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 57565 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0807, + "step": 57566 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 57567 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0881, + "step": 57568 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0873, + "step": 57569 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 57570 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0658, + "step": 57571 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 57572 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 57573 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 57574 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0826, + "step": 57575 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 57576 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 57577 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 57578 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 57579 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 57580 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 57581 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 57582 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 57583 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 57584 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 57585 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 57586 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0848, + "step": 57587 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.084, + "step": 57588 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 57589 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 57590 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 57591 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 57592 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 57593 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 57594 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 57595 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 57596 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 57597 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 57598 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 57599 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 57600 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0656, + "step": 57601 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 57602 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 57603 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0822, + "step": 57604 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.136, + "step": 57605 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.08, + "step": 57606 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0747, + "step": 57607 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0865, + "step": 57608 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 57609 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0952, + "step": 57610 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 57611 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0881, + "step": 57612 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0841, + "step": 57613 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 57614 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 57615 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0857, + "step": 57616 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 57617 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 57618 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 57619 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0754, + "step": 57620 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 57621 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 57622 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 57623 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0831, + "step": 57624 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 57625 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 57626 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 57627 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1355, + "step": 57628 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 57629 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0897, + "step": 57630 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 57631 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0835, + "step": 57632 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1301, + "step": 57633 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 57634 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 57635 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 57636 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 57637 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 57638 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1343, + "step": 57639 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 57640 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 57641 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 57642 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 57643 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 57644 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 57645 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0793, + "step": 57646 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 57647 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 57648 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 57649 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 57650 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 57651 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 57652 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0883, + "step": 57653 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 57654 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 57655 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0871, + "step": 57656 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0819, + "step": 57657 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 57658 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 57659 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0753, + "step": 57660 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 57661 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 57662 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 57663 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 57664 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 57665 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 57666 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 57667 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 57668 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 57669 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 57670 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 57671 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 57672 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 57673 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 57674 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 57675 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 57676 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 57677 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 57678 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0802, + "step": 57679 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 57680 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0728, + "step": 57681 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 57682 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0883, + "step": 57683 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 57684 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 57685 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 57686 + }, + { + "epoch": 3.5, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 57687 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0715, + "step": 57688 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 57689 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1368, + "step": 57690 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0789, + "step": 57691 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0967, + "step": 57692 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 57693 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 57694 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0863, + "step": 57695 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 57696 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0864, + "step": 57697 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 57698 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 57699 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0879, + "step": 57700 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0827, + "step": 57701 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 57702 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1599, + "step": 57703 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 57704 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 57705 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0877, + "step": 57706 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 57707 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0798, + "step": 57708 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1276, + "step": 57709 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0873, + "step": 57710 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 57711 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 57712 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 57713 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 57714 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 57715 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 57716 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0794, + "step": 57717 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 57718 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 57719 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 57720 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0915, + "step": 57721 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 57722 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 57723 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 57724 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0785, + "step": 57725 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 57726 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 57727 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0865, + "step": 57728 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0793, + "step": 57729 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 57730 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 57731 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 57732 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 57733 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0827, + "step": 57734 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 57735 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 57736 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0791, + "step": 57737 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0845, + "step": 57738 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0891, + "step": 57739 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0871, + "step": 57740 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 57741 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0823, + "step": 57742 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 57743 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 57744 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 57745 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 57746 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 57747 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0876, + "step": 57748 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 57749 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 57750 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0929, + "step": 57751 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.089, + "step": 57752 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 57753 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 57754 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 57755 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 57756 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0797, + "step": 57757 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 57758 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 57759 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0998, + "step": 57760 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0802, + "step": 57761 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 57762 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 57763 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0714, + "step": 57764 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0883, + "step": 57765 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 57766 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0875, + "step": 57767 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 57768 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 57769 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0857, + "step": 57770 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 57771 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 57772 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 57773 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 57774 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 57775 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0706, + "step": 57776 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 57777 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 57778 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 57779 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 57780 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 57781 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0895, + "step": 57782 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0802, + "step": 57783 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1407, + "step": 57784 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 57785 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 57786 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0937, + "step": 57787 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 57788 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 57789 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 57790 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 57791 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 57792 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 57793 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.145, + "step": 57794 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 57795 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 57796 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 57797 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 57798 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 57799 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0855, + "step": 57800 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.125, + "step": 57801 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 57802 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 57803 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 57804 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 57805 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0915, + "step": 57806 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0909, + "step": 57807 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 57808 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 57809 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 57810 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0842, + "step": 57811 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 57812 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 57813 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 57814 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 57815 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 57816 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 57817 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0878, + "step": 57818 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 57819 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1271, + "step": 57820 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 57821 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0952, + "step": 57822 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 57823 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 57824 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1333, + "step": 57825 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 57826 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 57827 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 57828 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 57829 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 57830 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 57831 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 57832 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 57833 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 57834 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 57835 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 57836 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 57837 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 57838 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.077, + "step": 57839 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 57840 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 57841 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 57842 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0787, + "step": 57843 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 57844 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 57845 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 57846 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0804, + "step": 57847 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 57848 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0876, + "step": 57849 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 57850 + }, + { + "epoch": 3.51, + "learning_rate": 2e-05, + "loss": 0.0866, + "step": 57851 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0858, + "step": 57852 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0747, + "step": 57853 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 57854 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.072, + "step": 57855 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 57856 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 57857 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0852, + "step": 57858 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 57859 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 57860 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 57861 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 57862 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 57863 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0935, + "step": 57864 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0875, + "step": 57865 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 57866 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.079, + "step": 57867 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 57868 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 57869 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1439, + "step": 57870 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0906, + "step": 57871 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 57872 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 57873 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 57874 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1289, + "step": 57875 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.079, + "step": 57876 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 57877 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0787, + "step": 57878 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0786, + "step": 57879 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 57880 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 57881 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 57882 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 57883 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 57884 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 57885 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 57886 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0898, + "step": 57887 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 57888 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 57889 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 57890 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 57891 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 57892 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 57893 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 57894 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 57895 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 57896 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0911, + "step": 57897 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0918, + "step": 57898 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 57899 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 57900 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 57901 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 57902 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 57903 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.08, + "step": 57904 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 57905 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 57906 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 57907 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0703, + "step": 57908 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 57909 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 57910 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 57911 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 57912 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 57913 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1125, + "step": 57914 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 57915 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 57916 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0875, + "step": 57917 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 57918 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0865, + "step": 57919 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.085, + "step": 57920 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 57921 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 57922 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 57923 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0857, + "step": 57924 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 57925 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 57926 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 57927 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 57928 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 57929 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 57930 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 57931 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 57932 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 57933 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 57934 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 57935 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 57936 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0845, + "step": 57937 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0878, + "step": 57938 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0689, + "step": 57939 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 57940 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 57941 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 57942 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0908, + "step": 57943 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 57944 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0843, + "step": 57945 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 57946 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 57947 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 57948 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0876, + "step": 57949 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 57950 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.149, + "step": 57951 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 57952 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 57953 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 57954 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1174, + "step": 57955 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 57956 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 57957 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 57958 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0849, + "step": 57959 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0838, + "step": 57960 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.081, + "step": 57961 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 57962 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 57963 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 57964 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0909, + "step": 57965 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 57966 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 57967 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 57968 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 57969 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 57970 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 57971 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 57972 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 57973 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0867, + "step": 57974 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0867, + "step": 57975 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0881, + "step": 57976 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 57977 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0911, + "step": 57978 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 57979 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 57980 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0839, + "step": 57981 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 57982 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 57983 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0832, + "step": 57984 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 57985 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0847, + "step": 57986 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 57987 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0865, + "step": 57988 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0898, + "step": 57989 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 57990 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 57991 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 57992 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 57993 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0941, + "step": 57994 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 57995 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 57996 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1317, + "step": 57997 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 57998 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 57999 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 58000 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 58001 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 58002 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1206, + "step": 58003 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0815, + "step": 58004 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0886, + "step": 58005 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 58006 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 58007 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 58008 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 58009 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0757, + "step": 58010 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1401, + "step": 58011 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 58012 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 58013 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1237, + "step": 58014 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 58015 + }, + { + "epoch": 3.52, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 58016 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0816, + "step": 58017 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 58018 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0788, + "step": 58019 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 58020 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 58021 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0918, + "step": 58022 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 58023 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 58024 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0848, + "step": 58025 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 58026 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 58027 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 58028 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 58029 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0902, + "step": 58030 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0897, + "step": 58031 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0853, + "step": 58032 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 58033 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 58034 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 58035 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 58036 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0851, + "step": 58037 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 58038 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 58039 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 58040 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 58041 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0856, + "step": 58042 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 58043 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0895, + "step": 58044 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1316, + "step": 58045 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 58046 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0869, + "step": 58047 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 58048 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 58049 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0867, + "step": 58050 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.085, + "step": 58051 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 58052 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 58053 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 58054 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 58055 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0746, + "step": 58056 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 58057 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 58058 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 58059 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 58060 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 58061 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 58062 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 58063 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.088, + "step": 58064 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 58065 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 58066 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 58067 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0858, + "step": 58068 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 58069 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.2167, + "step": 58070 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0754, + "step": 58071 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 58072 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 58073 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 58074 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 58075 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 58076 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 58077 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 58078 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 58079 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0855, + "step": 58080 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1383, + "step": 58081 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 58082 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 58083 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.075, + "step": 58084 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0891, + "step": 58085 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 58086 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0839, + "step": 58087 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0952, + "step": 58088 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 58089 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 58090 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 58091 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 58092 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0786, + "step": 58093 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 58094 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0883, + "step": 58095 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 58096 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 58097 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0722, + "step": 58098 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 58099 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 58100 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 58101 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.087, + "step": 58102 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.082, + "step": 58103 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 58104 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0899, + "step": 58105 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 58106 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0882, + "step": 58107 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0836, + "step": 58108 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 58109 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 58110 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0935, + "step": 58111 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 58112 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 58113 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 58114 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0828, + "step": 58115 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 58116 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 58117 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0828, + "step": 58118 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 58119 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0795, + "step": 58120 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 58121 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 58122 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 58123 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 58124 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 58125 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 58126 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 58127 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 58128 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 58129 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 58130 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 58131 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 58132 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 58133 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 58134 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 58135 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 58136 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0671, + "step": 58137 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0803, + "step": 58138 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 58139 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 58140 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 58141 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 58142 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0821, + "step": 58143 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 58144 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0837, + "step": 58145 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 58146 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 58147 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 58148 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 58149 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 58150 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 58151 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 58152 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 58153 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 58154 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 58155 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 58156 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0916, + "step": 58157 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 58158 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 58159 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 58160 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0914, + "step": 58161 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0865, + "step": 58162 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 58163 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0909, + "step": 58164 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0864, + "step": 58165 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 58166 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 58167 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 58168 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 58169 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 58170 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 58171 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.09, + "step": 58172 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 58173 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 58174 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 58175 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 58176 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 58177 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 58178 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 58179 + }, + { + "epoch": 3.53, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 58180 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 58181 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 58182 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 58183 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0894, + "step": 58184 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 58185 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 58186 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 58187 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 58188 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 58189 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 58190 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 58191 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 58192 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 58193 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 58194 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0932, + "step": 58195 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 58196 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 58197 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 58198 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 58199 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1396, + "step": 58200 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 58201 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1329, + "step": 58202 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 58203 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 58204 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0915, + "step": 58205 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0801, + "step": 58206 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 58207 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0709, + "step": 58208 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 58209 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 58210 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 58211 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 58212 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 58213 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 58214 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 58215 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0936, + "step": 58216 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 58217 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 58218 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 58219 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0754, + "step": 58220 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0836, + "step": 58221 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 58222 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 58223 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 58224 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 58225 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 58226 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 58227 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 58228 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 58229 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1351, + "step": 58230 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 58231 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 58232 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0875, + "step": 58233 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 58234 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0806, + "step": 58235 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 58236 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 58237 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 58238 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 58239 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.091, + "step": 58240 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 58241 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 58242 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 58243 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 58244 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0776, + "step": 58245 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 58246 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 58247 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0833, + "step": 58248 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 58249 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0846, + "step": 58250 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 58251 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0833, + "step": 58252 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 58253 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 58254 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1369, + "step": 58255 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1359, + "step": 58256 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0918, + "step": 58257 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 58258 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 58259 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 58260 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 58261 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 58262 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 58263 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.087, + "step": 58264 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 58265 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 58266 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 58267 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 58268 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 58269 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 58270 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0916, + "step": 58271 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0915, + "step": 58272 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0839, + "step": 58273 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 58274 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 58275 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 58276 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0943, + "step": 58277 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0745, + "step": 58278 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 58279 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0811, + "step": 58280 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 58281 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 58282 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 58283 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 58284 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 58285 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 58286 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1335, + "step": 58287 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 58288 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0903, + "step": 58289 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1577, + "step": 58290 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 58291 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 58292 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 58293 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 58294 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0929, + "step": 58295 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 58296 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 58297 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0866, + "step": 58298 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 58299 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0849, + "step": 58300 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 58301 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 58302 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0738, + "step": 58303 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 58304 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 58305 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0916, + "step": 58306 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 58307 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0876, + "step": 58308 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 58309 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 58310 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 58311 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 58312 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 58313 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 58314 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0859, + "step": 58315 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 58316 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 58317 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 58318 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.087, + "step": 58319 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 58320 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0832, + "step": 58321 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0835, + "step": 58322 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 58323 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 58324 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 58325 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0769, + "step": 58326 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 58327 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 58328 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0824, + "step": 58329 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0771, + "step": 58330 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0849, + "step": 58331 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 58332 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0856, + "step": 58333 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 58334 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 58335 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 58336 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 58337 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 58338 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 58339 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0789, + "step": 58340 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 58341 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 58342 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 58343 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.137, + "step": 58344 + }, + { + "epoch": 3.54, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 58345 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 58346 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 58347 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1392, + "step": 58348 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 58349 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0906, + "step": 58350 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0874, + "step": 58351 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 58352 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 58353 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0794, + "step": 58354 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 58355 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 58356 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0848, + "step": 58357 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 58358 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0874, + "step": 58359 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 58360 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 58361 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1279, + "step": 58362 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1432, + "step": 58363 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 58364 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 58365 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 58366 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 58367 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 58368 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0847, + "step": 58369 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 58370 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 58371 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 58372 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 58373 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 58374 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1379, + "step": 58375 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0953, + "step": 58376 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 58377 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 58378 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0772, + "step": 58379 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 58380 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 58381 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 58382 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 58383 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 58384 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0866, + "step": 58385 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 58386 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0874, + "step": 58387 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 58388 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 58389 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 58390 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 58391 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0796, + "step": 58392 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 58393 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 58394 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 58395 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 58396 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0767, + "step": 58397 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 58398 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0814, + "step": 58399 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 58400 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 58401 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 58402 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 58403 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 58404 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0914, + "step": 58405 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 58406 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 58407 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 58408 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0881, + "step": 58409 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0714, + "step": 58410 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 58411 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 58412 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0736, + "step": 58413 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0855, + "step": 58414 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 58415 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0898, + "step": 58416 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 58417 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 58418 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 58419 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 58420 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 58421 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1384, + "step": 58422 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.076, + "step": 58423 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 58424 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 58425 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 58426 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 58427 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 58428 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 58429 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 58430 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0832, + "step": 58431 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0863, + "step": 58432 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0878, + "step": 58433 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 58434 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 58435 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 58436 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 58437 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1176, + "step": 58438 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 58439 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 58440 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 58441 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 58442 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1406, + "step": 58443 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0789, + "step": 58444 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0932, + "step": 58445 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 58446 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.084, + "step": 58447 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 58448 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0763, + "step": 58449 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 58450 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 58451 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 58452 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 58453 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 58454 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0792, + "step": 58455 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 58456 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 58457 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0806, + "step": 58458 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 58459 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 58460 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0793, + "step": 58461 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 58462 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 58463 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 58464 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 58465 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0903, + "step": 58466 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 58467 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 58468 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 58469 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.076, + "step": 58470 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0856, + "step": 58471 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 58472 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 58473 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 58474 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1269, + "step": 58475 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 58476 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0784, + "step": 58477 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1197, + "step": 58478 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 58479 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 58480 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0837, + "step": 58481 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0875, + "step": 58482 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.087, + "step": 58483 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0828, + "step": 58484 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 58485 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0762, + "step": 58486 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 58487 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 58488 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 58489 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0668, + "step": 58490 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0821, + "step": 58491 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0788, + "step": 58492 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 58493 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0725, + "step": 58494 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 58495 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 58496 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 58497 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 58498 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0902, + "step": 58499 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 58500 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 58501 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 58502 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 58503 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 58504 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 58505 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 58506 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1161, + "step": 58507 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1246, + "step": 58508 + }, + { + "epoch": 3.55, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 58509 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1196, + "step": 58510 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 58511 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1378, + "step": 58512 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 58513 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 58514 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0768, + "step": 58515 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 58516 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 58517 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 58518 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 58519 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 58520 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 58521 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 58522 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 58523 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 58524 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0685, + "step": 58525 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 58526 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 58527 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 58528 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 58529 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 58530 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 58531 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0897, + "step": 58532 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0714, + "step": 58533 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1143, + "step": 58534 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 58535 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 58536 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0883, + "step": 58537 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 58538 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 58539 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 58540 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1011, + "step": 58541 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 58542 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0712, + "step": 58543 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 58544 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1112, + "step": 58545 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 58546 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 58547 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 58548 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1421, + "step": 58549 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 58550 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0838, + "step": 58551 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0842, + "step": 58552 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0899, + "step": 58553 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 58554 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 58555 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 58556 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.096, + "step": 58557 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 58558 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0872, + "step": 58559 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1879, + "step": 58560 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 58561 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 58562 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.084, + "step": 58563 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 58564 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 58565 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 58566 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 58567 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 58568 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 58569 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0859, + "step": 58570 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 58571 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 58572 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 58573 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.141, + "step": 58574 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 58575 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 58576 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 58577 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 58578 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0897, + "step": 58579 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0817, + "step": 58580 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 58581 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0898, + "step": 58582 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 58583 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 58584 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0718, + "step": 58585 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0875, + "step": 58586 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 58587 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 58588 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1139, + "step": 58589 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0914, + "step": 58590 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 58591 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 58592 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 58593 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0864, + "step": 58594 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 58595 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 58596 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 58597 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1244, + "step": 58598 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 58599 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 58600 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 58601 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 58602 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0883, + "step": 58603 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 58604 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 58605 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0935, + "step": 58606 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 58607 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 58608 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 58609 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 58610 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 58611 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 58612 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 58613 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0906, + "step": 58614 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1261, + "step": 58615 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1473, + "step": 58616 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1267, + "step": 58617 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 58618 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 58619 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0951, + "step": 58620 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 58621 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 58622 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1418, + "step": 58623 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 58624 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 58625 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 58626 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1062, + "step": 58627 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 58628 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0788, + "step": 58629 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 58630 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 58631 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 58632 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 58633 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 58634 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 58635 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 58636 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 58637 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 58638 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 58639 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 58640 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 58641 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0855, + "step": 58642 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 58643 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0745, + "step": 58644 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0866, + "step": 58645 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 58646 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 58647 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 58648 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 58649 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 58650 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 58651 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 58652 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 58653 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1031, + "step": 58654 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 58655 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 58656 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 58657 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 58658 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 58659 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 58660 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1096, + "step": 58661 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1551, + "step": 58662 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 58663 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 58664 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0891, + "step": 58665 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 58666 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0829, + "step": 58667 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.076, + "step": 58668 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0725, + "step": 58669 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 58670 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 58671 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 58672 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.0836, + "step": 58673 + }, + { + "epoch": 3.56, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 58674 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 58675 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 58676 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 58677 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 58678 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 58679 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0802, + "step": 58680 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0698, + "step": 58681 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 58682 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 58683 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 58684 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 58685 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0946, + "step": 58686 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 58687 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 58688 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1264, + "step": 58689 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.088, + "step": 58690 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 58691 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 58692 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.081, + "step": 58693 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 58694 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 58695 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0848, + "step": 58696 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0887, + "step": 58697 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 58698 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 58699 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.074, + "step": 58700 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0823, + "step": 58701 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1192, + "step": 58702 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 58703 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0855, + "step": 58704 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 58705 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 58706 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 58707 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 58708 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0898, + "step": 58709 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 58710 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1171, + "step": 58711 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 58712 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 58713 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1186, + "step": 58714 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 58715 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 58716 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 58717 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 58718 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.113, + "step": 58719 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 58720 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 58721 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 58722 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 58723 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 58724 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 58725 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 58726 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 58727 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 58728 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0656, + "step": 58729 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 58730 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0879, + "step": 58731 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 58732 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 58733 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 58734 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 58735 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 58736 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 58737 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 58738 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 58739 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.088, + "step": 58740 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1215, + "step": 58741 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 58742 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0836, + "step": 58743 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 58744 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 58745 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0813, + "step": 58746 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 58747 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 58748 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 58749 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 58750 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 58751 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 58752 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1367, + "step": 58753 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 58754 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 58755 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0808, + "step": 58756 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 58757 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 58758 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1243, + "step": 58759 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 58760 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 58761 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.071, + "step": 58762 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 58763 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 58764 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 58765 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 58766 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 58767 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 58768 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 58769 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0784, + "step": 58770 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 58771 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 58772 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 58773 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 58774 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 58775 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 58776 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 58777 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 58778 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 58779 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 58780 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 58781 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 58782 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0882, + "step": 58783 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 58784 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 58785 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 58786 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 58787 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0805, + "step": 58788 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 58789 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 58790 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 58791 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0674, + "step": 58792 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 58793 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 58794 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 58795 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0852, + "step": 58796 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 58797 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 58798 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0889, + "step": 58799 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 58800 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 58801 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 58802 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 58803 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.083, + "step": 58804 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 58805 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 58806 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1347, + "step": 58807 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 58808 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0879, + "step": 58809 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0883, + "step": 58810 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0901, + "step": 58811 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1386, + "step": 58812 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0813, + "step": 58813 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 58814 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 58815 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 58816 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 58817 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0777, + "step": 58818 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0858, + "step": 58819 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 58820 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 58821 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 58822 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 58823 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 58824 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 58825 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 58826 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 58827 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 58828 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 58829 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 58830 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 58831 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1416, + "step": 58832 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.0892, + "step": 58833 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 58834 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1221, + "step": 58835 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1155, + "step": 58836 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 58837 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 58838 + }, + { + "epoch": 3.57, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 58839 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1371, + "step": 58840 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 58841 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.119, + "step": 58842 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 58843 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.08, + "step": 58844 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 58845 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 58846 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 58847 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 58848 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0857, + "step": 58849 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 58850 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0817, + "step": 58851 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 58852 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 58853 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1275, + "step": 58854 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 58855 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0914, + "step": 58856 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0894, + "step": 58857 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 58858 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 58859 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 58860 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 58861 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 58862 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1303, + "step": 58863 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 58864 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 58865 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1204, + "step": 58866 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0849, + "step": 58867 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0901, + "step": 58868 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 58869 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 58870 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 58871 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1309, + "step": 58872 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0819, + "step": 58873 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 58874 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0679, + "step": 58875 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 58876 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0844, + "step": 58877 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 58878 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 58879 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 58880 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0875, + "step": 58881 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 58882 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 58883 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 58884 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 58885 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 58886 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 58887 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 58888 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 58889 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0929, + "step": 58890 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0862, + "step": 58891 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 58892 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 58893 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 58894 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 58895 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 58896 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1124, + "step": 58897 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 58898 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 58899 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1358, + "step": 58900 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1051, + "step": 58901 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 58902 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 58903 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 58904 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 58905 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 58906 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 58907 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 58908 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 58909 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 58910 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 58911 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 58912 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 58913 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0891, + "step": 58914 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 58915 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 58916 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0824, + "step": 58917 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 58918 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 58919 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 58920 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 58921 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 58922 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0842, + "step": 58923 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0952, + "step": 58924 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 58925 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 58926 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 58927 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 58928 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 58929 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.139, + "step": 58930 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 58931 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0813, + "step": 58932 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 58933 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 58934 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 58935 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 58936 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 58937 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1078, + "step": 58938 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 58939 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0738, + "step": 58940 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0816, + "step": 58941 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 58942 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 58943 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 58944 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1207, + "step": 58945 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.091, + "step": 58946 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1189, + "step": 58947 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 58948 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0902, + "step": 58949 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1334, + "step": 58950 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0842, + "step": 58951 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1184, + "step": 58952 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 58953 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0853, + "step": 58954 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 58955 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 58956 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 58957 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0865, + "step": 58958 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1412, + "step": 58959 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 58960 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0894, + "step": 58961 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.088, + "step": 58962 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 58963 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0903, + "step": 58964 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0772, + "step": 58965 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 58966 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 58967 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 58968 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 58969 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 58970 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0792, + "step": 58971 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 58972 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 58973 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0782, + "step": 58974 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 58975 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 58976 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0916, + "step": 58977 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 58978 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1436, + "step": 58979 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 58980 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 58981 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 58982 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.081, + "step": 58983 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 58984 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 58985 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 58986 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.092, + "step": 58987 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 58988 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 58989 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 58990 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0834, + "step": 58991 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 58992 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 58993 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 58994 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1068, + "step": 58995 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 58996 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1017, + "step": 58997 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1177, + "step": 58998 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0823, + "step": 58999 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 59000 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0897, + "step": 59001 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 59002 + }, + { + "epoch": 3.58, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 59003 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 59004 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 59005 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 59006 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 59007 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 59008 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 59009 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 59010 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0791, + "step": 59011 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0741, + "step": 59012 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1321, + "step": 59013 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0869, + "step": 59014 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1241, + "step": 59015 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0857, + "step": 59016 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 59017 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0873, + "step": 59018 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 59019 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 59020 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0932, + "step": 59021 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 59022 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0781, + "step": 59023 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1131, + "step": 59024 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 59025 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 59026 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 59027 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 59028 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1005, + "step": 59029 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0781, + "step": 59030 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 59031 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0847, + "step": 59032 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 59033 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 59034 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 59035 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 59036 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 59037 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 59038 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 59039 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.081, + "step": 59040 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 59041 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0914, + "step": 59042 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 59043 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 59044 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 59045 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0804, + "step": 59046 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1058, + "step": 59047 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 59048 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0804, + "step": 59049 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0868, + "step": 59050 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1079, + "step": 59051 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 59052 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 59053 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0916, + "step": 59054 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 59055 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 59056 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 59057 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 59058 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 59059 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 59060 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.155, + "step": 59061 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 59062 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 59063 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 59064 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 59065 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0939, + "step": 59066 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1353, + "step": 59067 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1252, + "step": 59068 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 59069 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 59070 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1345, + "step": 59071 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 59072 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 59073 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 59074 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 59075 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 59076 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0838, + "step": 59077 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 59078 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 59079 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0772, + "step": 59080 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0891, + "step": 59081 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 59082 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 59083 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 59084 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 59085 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 59086 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1251, + "step": 59087 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1129, + "step": 59088 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 59089 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 59090 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0914, + "step": 59091 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1739, + "step": 59092 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 59093 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 59094 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1217, + "step": 59095 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 59096 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 59097 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 59098 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 59099 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 59100 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 59101 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 59102 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1202, + "step": 59103 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1715, + "step": 59104 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 59105 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 59106 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 59107 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 59108 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 59109 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0932, + "step": 59110 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 59111 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0886, + "step": 59112 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 59113 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0855, + "step": 59114 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 59115 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 59116 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 59117 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 59118 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 59119 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0947, + "step": 59120 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0759, + "step": 59121 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 59122 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1266, + "step": 59123 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 59124 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 59125 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 59126 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0888, + "step": 59127 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 59128 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0901, + "step": 59129 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 59130 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 59131 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 59132 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.132, + "step": 59133 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0823, + "step": 59134 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0749, + "step": 59135 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0862, + "step": 59136 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0889, + "step": 59137 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 59138 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1138, + "step": 59139 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 59140 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 59141 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 59142 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 59143 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 59144 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 59145 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 59146 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1208, + "step": 59147 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0862, + "step": 59148 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 59149 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 59150 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0655, + "step": 59151 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.082, + "step": 59152 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 59153 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 59154 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 59155 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0895, + "step": 59156 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 59157 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 59158 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 59159 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 59160 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 59161 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1397, + "step": 59162 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0918, + "step": 59163 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 59164 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.087, + "step": 59165 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 59166 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 59167 + }, + { + "epoch": 3.59, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 59168 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 59169 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1026, + "step": 59170 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 59171 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0856, + "step": 59172 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 59173 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 59174 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 59175 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 59176 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0851, + "step": 59177 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 59178 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0894, + "step": 59179 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0837, + "step": 59180 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 59181 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.105, + "step": 59182 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 59183 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 59184 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0911, + "step": 59185 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0819, + "step": 59186 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 59187 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0849, + "step": 59188 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 59189 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 59190 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 59191 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 59192 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 59193 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1041, + "step": 59194 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1249, + "step": 59195 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 59196 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0847, + "step": 59197 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0792, + "step": 59198 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0784, + "step": 59199 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 59200 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 59201 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 59202 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0811, + "step": 59203 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0825, + "step": 59204 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 59205 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1027, + "step": 59206 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1061, + "step": 59207 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0894, + "step": 59208 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 59209 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1098, + "step": 59210 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0772, + "step": 59211 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1352, + "step": 59212 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 59213 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0863, + "step": 59214 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 59215 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0902, + "step": 59216 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0849, + "step": 59217 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 59218 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 59219 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 59220 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 59221 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 59222 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 59223 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0906, + "step": 59224 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 59225 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 59226 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0862, + "step": 59227 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1012, + "step": 59228 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1024, + "step": 59229 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 59230 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0932, + "step": 59231 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0864, + "step": 59232 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0863, + "step": 59233 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 59234 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 59235 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1226, + "step": 59236 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0999, + "step": 59237 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 59238 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 59239 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 59240 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0767, + "step": 59241 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 59242 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 59243 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 59244 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 59245 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.093, + "step": 59246 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 59247 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 59248 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 59249 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0917, + "step": 59250 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 59251 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 59252 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 59253 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 59254 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1388, + "step": 59255 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 59256 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 59257 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 59258 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 59259 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 59260 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 59261 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0895, + "step": 59262 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 59263 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 59264 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0764, + "step": 59265 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1212, + "step": 59266 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 59267 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1147, + "step": 59268 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 59269 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 59270 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 59271 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 59272 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 59273 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0661, + "step": 59274 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0938, + "step": 59275 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1308, + "step": 59276 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 59277 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0906, + "step": 59278 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0828, + "step": 59279 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 59280 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 59281 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 59282 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 59283 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 59284 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0929, + "step": 59285 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 59286 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 59287 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 59288 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 59289 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0891, + "step": 59290 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1075, + "step": 59291 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 59292 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0897, + "step": 59293 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0885, + "step": 59294 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 59295 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 59296 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 59297 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 59298 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 59299 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 59300 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 59301 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1346, + "step": 59302 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 59303 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0889, + "step": 59304 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 59305 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0846, + "step": 59306 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 59307 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 59308 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1166, + "step": 59309 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 59310 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 59311 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 59312 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 59313 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1332, + "step": 59314 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 59315 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0856, + "step": 59316 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 59317 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 59318 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0915, + "step": 59319 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 59320 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 59321 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1274, + "step": 59322 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0921, + "step": 59323 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0877, + "step": 59324 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1014, + "step": 59325 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 59326 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0775, + "step": 59327 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1006, + "step": 59328 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0904, + "step": 59329 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 59330 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 59331 + }, + { + "epoch": 3.6, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 59332 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 59333 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 59334 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1179, + "step": 59335 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0791, + "step": 59336 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 59337 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 59338 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1066, + "step": 59339 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1123, + "step": 59340 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 59341 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1195, + "step": 59342 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 59343 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0848, + "step": 59344 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 59345 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0794, + "step": 59346 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0842, + "step": 59347 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 59348 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1118, + "step": 59349 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 59350 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 59351 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1073, + "step": 59352 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 59353 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1283, + "step": 59354 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 59355 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1183, + "step": 59356 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 59357 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 59358 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0927, + "step": 59359 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 59360 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 59361 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 59362 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 59363 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0868, + "step": 59364 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0872, + "step": 59365 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 59366 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 59367 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 59368 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 59369 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 59370 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.072, + "step": 59371 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 59372 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1326, + "step": 59373 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 59374 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 59375 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1222, + "step": 59376 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0849, + "step": 59377 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 59378 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 59379 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 59380 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 59381 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1033, + "step": 59382 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0954, + "step": 59383 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 59384 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 59385 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 59386 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.117, + "step": 59387 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0915, + "step": 59388 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0903, + "step": 59389 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 59390 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 59391 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 59392 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.08, + "step": 59393 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0966, + "step": 59394 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0867, + "step": 59395 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 59396 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0934, + "step": 59397 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 59398 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0863, + "step": 59399 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1405, + "step": 59400 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 59401 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0841, + "step": 59402 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 59403 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0814, + "step": 59404 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 59405 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 59406 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 59407 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1336, + "step": 59408 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0806, + "step": 59409 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 59410 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0659, + "step": 59411 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 59412 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 59413 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0924, + "step": 59414 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1148, + "step": 59415 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0842, + "step": 59416 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0834, + "step": 59417 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0802, + "step": 59418 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1356, + "step": 59419 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0933, + "step": 59420 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1067, + "step": 59421 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0839, + "step": 59422 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1087, + "step": 59423 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1331, + "step": 59424 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0962, + "step": 59425 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0863, + "step": 59426 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 59427 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 59428 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 59429 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 59430 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 59431 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0869, + "step": 59432 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 59433 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 59434 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 59435 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 59436 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.099, + "step": 59437 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 59438 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0687, + "step": 59439 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 59440 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 59441 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 59442 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0949, + "step": 59443 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 59444 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 59445 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 59446 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0864, + "step": 59447 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 59448 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0892, + "step": 59449 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 59450 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 59451 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 59452 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1135, + "step": 59453 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1399, + "step": 59454 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 59455 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 59456 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0859, + "step": 59457 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1323, + "step": 59458 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0914, + "step": 59459 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 59460 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 59461 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0801, + "step": 59462 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0869, + "step": 59463 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 59464 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 59465 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 59466 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 59467 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 59468 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0925, + "step": 59469 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.103, + "step": 59470 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0797, + "step": 59471 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 59472 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 59473 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 59474 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 59475 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 59476 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0681, + "step": 59477 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0984, + "step": 59478 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 59479 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 59480 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1036, + "step": 59481 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0832, + "step": 59482 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 59483 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 59484 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1263, + "step": 59485 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0931, + "step": 59486 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1055, + "step": 59487 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 59488 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0727, + "step": 59489 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 59490 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1337, + "step": 59491 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0759, + "step": 59492 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0753, + "step": 59493 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 59494 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.088, + "step": 59495 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0964, + "step": 59496 + }, + { + "epoch": 3.61, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 59497 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.111, + "step": 59498 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 59499 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0836, + "step": 59500 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 59501 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0723, + "step": 59502 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1167, + "step": 59503 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 59504 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1445, + "step": 59505 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0883, + "step": 59506 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 59507 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1165, + "step": 59508 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 59509 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 59510 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 59511 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1084, + "step": 59512 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0872, + "step": 59513 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0891, + "step": 59514 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1169, + "step": 59515 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 59516 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 59517 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0981, + "step": 59518 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 59519 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 59520 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1236, + "step": 59521 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1298, + "step": 59522 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 59523 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 59524 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 59525 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 59526 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0841, + "step": 59527 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 59528 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0889, + "step": 59529 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 59530 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.086, + "step": 59531 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1137, + "step": 59532 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0905, + "step": 59533 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1158, + "step": 59534 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 59535 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 59536 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1095, + "step": 59537 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 59538 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0952, + "step": 59539 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 59540 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1013, + "step": 59541 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 59542 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0863, + "step": 59543 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0789, + "step": 59544 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 59545 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 59546 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0667, + "step": 59547 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 59548 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1254, + "step": 59549 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 59550 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 59551 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0971, + "step": 59552 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0995, + "step": 59553 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 59554 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1156, + "step": 59555 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 59556 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 59557 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1314, + "step": 59558 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 59559 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 59560 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0771, + "step": 59561 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.107, + "step": 59562 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1227, + "step": 59563 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1238, + "step": 59564 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0926, + "step": 59565 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1294, + "step": 59566 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1191, + "step": 59567 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1173, + "step": 59568 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 59569 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 59570 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 59571 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 59572 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1094, + "step": 59573 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1072, + "step": 59574 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.095, + "step": 59575 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0902, + "step": 59576 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 59577 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1004, + "step": 59578 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1292, + "step": 59579 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 59580 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1315, + "step": 59581 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1063, + "step": 59582 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1042, + "step": 59583 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 59584 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1248, + "step": 59585 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1374, + "step": 59586 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 59587 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 59588 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1037, + "step": 59589 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1115, + "step": 59590 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 59591 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 59592 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1141, + "step": 59593 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0745, + "step": 59594 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0865, + "step": 59595 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 59596 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 59597 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1233, + "step": 59598 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 59599 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1172, + "step": 59600 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0837, + "step": 59601 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0967, + "step": 59602 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1097, + "step": 59603 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0873, + "step": 59604 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.097, + "step": 59605 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 59606 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 59607 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0929, + "step": 59608 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.114, + "step": 59609 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 59610 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 59611 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0952, + "step": 59612 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 59613 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 59614 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0729, + "step": 59615 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1203, + "step": 59616 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0839, + "step": 59617 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 59618 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 59619 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 59620 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1362, + "step": 59621 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0744, + "step": 59622 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1085, + "step": 59623 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 59624 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 59625 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 59626 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0727, + "step": 59627 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 59628 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1018, + "step": 59629 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1235, + "step": 59630 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1007, + "step": 59631 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1198, + "step": 59632 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 59633 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0872, + "step": 59634 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 59635 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0707, + "step": 59636 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 59637 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 59638 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 59639 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0812, + "step": 59640 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0789, + "step": 59641 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.108, + "step": 59642 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 59643 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1306, + "step": 59644 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1119, + "step": 59645 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0907, + "step": 59646 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0979, + "step": 59647 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 59648 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 59649 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 59650 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1083, + "step": 59651 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1199, + "step": 59652 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0894, + "step": 59653 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1059, + "step": 59654 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 59655 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1106, + "step": 59656 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0701, + "step": 59657 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 59658 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.072, + "step": 59659 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 59660 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.1313, + "step": 59661 + }, + { + "epoch": 3.62, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 59662 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1136, + "step": 59663 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 59664 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 59665 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 59666 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 59667 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1082, + "step": 59668 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 59669 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0863, + "step": 59670 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0654, + "step": 59671 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1245, + "step": 59672 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 59673 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0782, + "step": 59674 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.077, + "step": 59675 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0972, + "step": 59676 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 59677 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0867, + "step": 59678 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0944, + "step": 59679 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1088, + "step": 59680 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 59681 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 59682 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0708, + "step": 59683 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 59684 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 59685 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 59686 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1151, + "step": 59687 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1548, + "step": 59688 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 59689 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 59690 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0932, + "step": 59691 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 59692 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0888, + "step": 59693 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 59694 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 59695 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 59696 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 59697 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1284, + "step": 59698 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0958, + "step": 59699 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 59700 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.129, + "step": 59701 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 59702 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 59703 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 59704 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0952, + "step": 59705 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0996, + "step": 59706 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1153, + "step": 59707 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1273, + "step": 59708 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 59709 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0997, + "step": 59710 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 59711 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0819, + "step": 59712 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0967, + "step": 59713 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1121, + "step": 59714 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 59715 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0895, + "step": 59716 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 59717 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1223, + "step": 59718 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 59719 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0872, + "step": 59720 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1047, + "step": 59721 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 59722 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1214, + "step": 59723 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0809, + "step": 59724 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 59725 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1218, + "step": 59726 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0861, + "step": 59727 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0976, + "step": 59728 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1122, + "step": 59729 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0942, + "step": 59730 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 59731 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 59732 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 59733 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1454, + "step": 59734 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0991, + "step": 59735 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 59736 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0785, + "step": 59737 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 59738 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 59739 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1039, + "step": 59740 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0912, + "step": 59741 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 59742 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0844, + "step": 59743 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0762, + "step": 59744 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1052, + "step": 59745 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0892, + "step": 59746 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 59747 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 59748 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 59749 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 59750 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 59751 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 59752 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1539, + "step": 59753 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 59754 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1057, + "step": 59755 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0916, + "step": 59756 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.121, + "step": 59757 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1142, + "step": 59758 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 59759 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0986, + "step": 59760 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1023, + "step": 59761 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0945, + "step": 59762 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1164, + "step": 59763 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 59764 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1193, + "step": 59765 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 59766 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1262, + "step": 59767 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 59768 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 59769 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1257, + "step": 59770 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.094, + "step": 59771 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1305, + "step": 59772 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0865, + "step": 59773 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 59774 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 59775 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1268, + "step": 59776 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1034, + "step": 59777 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 59778 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 59779 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 59780 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1028, + "step": 59781 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 59782 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0922, + "step": 59783 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 59784 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 59785 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 59786 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1213, + "step": 59787 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1128, + "step": 59788 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0899, + "step": 59789 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1032, + "step": 59790 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.106, + "step": 59791 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 59792 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0877, + "step": 59793 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0766, + "step": 59794 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0743, + "step": 59795 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0988, + "step": 59796 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1, + "step": 59797 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1049, + "step": 59798 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1324, + "step": 59799 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1307, + "step": 59800 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.09, + "step": 59801 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1163, + "step": 59802 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 59803 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 59804 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1101, + "step": 59805 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1029, + "step": 59806 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1441, + "step": 59807 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0992, + "step": 59808 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0765, + "step": 59809 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0873, + "step": 59810 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0911, + "step": 59811 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 59812 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1107, + "step": 59813 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 59814 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1431, + "step": 59815 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0916, + "step": 59816 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 59817 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0887, + "step": 59818 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1091, + "step": 59819 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1157, + "step": 59820 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.128, + "step": 59821 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1216, + "step": 59822 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 59823 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 59824 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.1168, + "step": 59825 + }, + { + "epoch": 3.63, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 59826 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 59827 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 59828 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0859, + "step": 59829 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0678, + "step": 59830 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0767, + "step": 59831 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1361, + "step": 59832 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0839, + "step": 59833 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1022, + "step": 59834 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0886, + "step": 59835 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1044, + "step": 59836 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 59837 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1339, + "step": 59838 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1259, + "step": 59839 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0839, + "step": 59840 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 59841 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1081, + "step": 59842 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.126, + "step": 59843 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1209, + "step": 59844 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1149, + "step": 59845 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1109, + "step": 59846 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 59847 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.085, + "step": 59848 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1045, + "step": 59849 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1288, + "step": 59850 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0895, + "step": 59851 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1103, + "step": 59852 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1035, + "step": 59853 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1285, + "step": 59854 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 59855 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0961, + "step": 59856 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0989, + "step": 59857 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0879, + "step": 59858 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1194, + "step": 59859 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1048, + "step": 59860 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0941, + "step": 59861 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1302, + "step": 59862 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0763, + "step": 59863 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1093, + "step": 59864 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0796, + "step": 59865 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1438, + "step": 59866 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1211, + "step": 59867 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 59868 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.091, + "step": 59869 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0705, + "step": 59870 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0864, + "step": 59871 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0818, + "step": 59872 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.098, + "step": 59873 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.116, + "step": 59874 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 59875 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.102, + "step": 59876 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0948, + "step": 59877 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0977, + "step": 59878 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0818, + "step": 59879 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1247, + "step": 59880 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.12, + "step": 59881 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.084, + "step": 59882 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0786, + "step": 59883 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1134, + "step": 59884 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1178, + "step": 59885 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0858, + "step": 59886 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1009, + "step": 59887 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0855, + "step": 59888 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0955, + "step": 59889 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 59890 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1105, + "step": 59891 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1053, + "step": 59892 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0956, + "step": 59893 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.082, + "step": 59894 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0993, + "step": 59895 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 59896 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1086, + "step": 59897 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1038, + "step": 59898 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1076, + "step": 59899 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.063, + "step": 59900 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1255, + "step": 59901 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0866, + "step": 59902 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0973, + "step": 59903 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1182, + "step": 59904 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 59905 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1111, + "step": 59906 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1104, + "step": 59907 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0957, + "step": 59908 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0862, + "step": 59909 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 59910 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1046, + "step": 59911 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0929, + "step": 59912 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0985, + "step": 59913 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0974, + "step": 59914 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 59915 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 59916 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 59917 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 59918 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1002, + "step": 59919 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0884, + "step": 59920 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1232, + "step": 59921 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1296, + "step": 59922 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1019, + "step": 59923 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0906, + "step": 59924 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0882, + "step": 59925 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1099, + "step": 59926 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1162, + "step": 59927 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.11, + "step": 59928 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0967, + "step": 59929 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0987, + "step": 59930 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 59931 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 59932 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1008, + "step": 59933 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1188, + "step": 59934 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.109, + "step": 59935 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0994, + "step": 59936 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0758, + "step": 59937 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1064, + "step": 59938 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1152, + "step": 59939 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.118, + "step": 59940 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 59941 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0963, + "step": 59942 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.101, + "step": 59943 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1185, + "step": 59944 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1074, + "step": 59945 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0886, + "step": 59946 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0975, + "step": 59947 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0959, + "step": 59948 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.112, + "step": 59949 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1065, + "step": 59950 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0982, + "step": 59951 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 59952 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.123, + "step": 59953 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1092, + "step": 59954 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1113, + "step": 59955 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 59956 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0882, + "step": 59957 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1015, + "step": 59958 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 59959 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0829, + "step": 59960 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1116, + "step": 59961 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1187, + "step": 59962 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0969, + "step": 59963 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1117, + "step": 59964 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1181, + "step": 59965 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0983, + "step": 59966 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1069, + "step": 59967 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.13, + "step": 59968 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0909, + "step": 59969 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.134, + "step": 59970 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1089, + "step": 59971 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 59972 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0893, + "step": 59973 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1126, + "step": 59974 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0603, + "step": 59975 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0978, + "step": 59976 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1056, + "step": 59977 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.122, + "step": 59978 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1132, + "step": 59979 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.104, + "step": 59980 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.0913, + "step": 59981 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1133, + "step": 59982 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1175, + "step": 59983 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1357, + "step": 59984 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1043, + "step": 59985 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1144, + "step": 59986 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 59987 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1025, + "step": 59988 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1108, + "step": 59989 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1299, + "step": 59990 + }, + { + "epoch": 3.64, + "learning_rate": 2e-05, + "loss": 0.1114, + "step": 59991 + }, + { + "epoch": 3.65, + "learning_rate": 2e-05, + "loss": 0.0968, + "step": 59992 + }, + { + "epoch": 3.65, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 59993 + }, + { + "epoch": 3.65, + "learning_rate": 2e-05, + "loss": 0.0923, + "step": 59994 + }, + { + "epoch": 3.65, + "learning_rate": 2e-05, + "loss": 0.1054, + "step": 59995 + }, + { + "epoch": 3.65, + "learning_rate": 2e-05, + "loss": 0.1102, + "step": 59996 + }, + { + "epoch": 3.65, + "learning_rate": 2e-05, + "loss": 0.1016, + "step": 59997 + }, + { + "epoch": 3.65, + "learning_rate": 2e-05, + "loss": 0.1071, + "step": 59998 + }, + { + "epoch": 3.65, + "learning_rate": 2e-05, + "loss": 0.1003, + "step": 59999 + }, + { + "epoch": 3.65, + "learning_rate": 2e-05, + "loss": 0.1127, + "step": 60000 + } + ], + "logging_steps": 1, + "max_steps": 82290, + "num_train_epochs": 5, + "save_steps": 500, + "total_flos": 2.8669068494077133e+19, + "trial_name": null, + "trial_params": null +}