diff --git "a/checkpoint-1838/trainer_state.json" "b/checkpoint-1838/trainer_state.json" new file mode 100644--- /dev/null +++ "b/checkpoint-1838/trainer_state.json" @@ -0,0 +1,11047 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 2.0, + "eval_steps": 500, + "global_step": 1838, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0, + "learning_rate": 2.777777777777778e-06, + "loss": 3.1281, + "step": 1 + }, + { + "epoch": 0.0, + "learning_rate": 5.555555555555556e-06, + "loss": 3.2134, + "step": 2 + }, + { + "epoch": 0.0, + "learning_rate": 8.333333333333334e-06, + "loss": 3.1329, + "step": 3 + }, + { + "epoch": 0.0, + "learning_rate": 1.1111111111111112e-05, + "loss": 3.1304, + "step": 4 + }, + { + "epoch": 0.01, + "learning_rate": 1.3888888888888888e-05, + "loss": 3.1683, + "step": 5 + }, + { + "epoch": 0.01, + "learning_rate": 1.6666666666666667e-05, + "loss": 3.2336, + "step": 6 + }, + { + "epoch": 0.01, + "learning_rate": 1.9444444444444445e-05, + "loss": 3.2424, + "step": 7 + }, + { + "epoch": 0.01, + "learning_rate": 2.2222222222222223e-05, + "loss": 3.2155, + "step": 8 + }, + { + "epoch": 0.01, + "learning_rate": 2.5e-05, + "loss": 3.1961, + "step": 9 + }, + { + "epoch": 0.01, + "learning_rate": 2.7777777777777776e-05, + "loss": 3.2881, + "step": 10 + }, + { + "epoch": 0.01, + "learning_rate": 3.0555555555555554e-05, + "loss": 3.1787, + "step": 11 + }, + { + "epoch": 0.01, + "learning_rate": 3.3333333333333335e-05, + "loss": 3.2492, + "step": 12 + }, + { + "epoch": 0.01, + "learning_rate": 3.611111111111111e-05, + "loss": 3.0484, + "step": 13 + }, + { + "epoch": 0.02, + "learning_rate": 3.888888888888889e-05, + "loss": 3.0527, + "step": 14 + }, + { + "epoch": 0.02, + "learning_rate": 4.1666666666666665e-05, + "loss": 3.1198, + "step": 15 + }, + { + "epoch": 0.02, + "learning_rate": 4.4444444444444447e-05, + "loss": 3.0216, + "step": 16 + }, + { + "epoch": 0.02, + "learning_rate": 4.722222222222222e-05, + "loss": 2.9886, + "step": 17 + }, + { + "epoch": 0.02, + "learning_rate": 5e-05, + "loss": 2.9654, + "step": 18 + }, + { + "epoch": 0.02, + "learning_rate": 5.277777777777778e-05, + "loss": 2.9086, + "step": 19 + }, + { + "epoch": 0.02, + "learning_rate": 5.555555555555555e-05, + "loss": 2.948, + "step": 20 + }, + { + "epoch": 0.02, + "learning_rate": 5.833333333333333e-05, + "loss": 2.9377, + "step": 21 + }, + { + "epoch": 0.02, + "learning_rate": 6.111111111111111e-05, + "loss": 2.9113, + "step": 22 + }, + { + "epoch": 0.03, + "learning_rate": 6.388888888888888e-05, + "loss": 2.7984, + "step": 23 + }, + { + "epoch": 0.03, + "learning_rate": 6.666666666666667e-05, + "loss": 2.7287, + "step": 24 + }, + { + "epoch": 0.03, + "learning_rate": 6.944444444444444e-05, + "loss": 2.6983, + "step": 25 + }, + { + "epoch": 0.03, + "learning_rate": 7.222222222222222e-05, + "loss": 2.6575, + "step": 26 + }, + { + "epoch": 0.03, + "learning_rate": 7.5e-05, + "loss": 2.6286, + "step": 27 + }, + { + "epoch": 0.03, + "learning_rate": 7.777777777777778e-05, + "loss": 2.5648, + "step": 28 + }, + { + "epoch": 0.03, + "learning_rate": 8.055555555555556e-05, + "loss": 2.5435, + "step": 29 + }, + { + "epoch": 0.03, + "learning_rate": 8.333333333333333e-05, + "loss": 2.5417, + "step": 30 + }, + { + "epoch": 0.03, + "learning_rate": 8.611111111111112e-05, + "loss": 2.4647, + "step": 31 + }, + { + "epoch": 0.03, + "learning_rate": 8.888888888888889e-05, + "loss": 2.4418, + "step": 32 + }, + { + "epoch": 0.04, + "learning_rate": 9.166666666666667e-05, + "loss": 2.3412, + "step": 33 + }, + { + "epoch": 0.04, + "learning_rate": 9.444444444444444e-05, + "loss": 2.3532, + "step": 34 + }, + { + "epoch": 0.04, + "learning_rate": 9.722222222222223e-05, + "loss": 2.2995, + "step": 35 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.3202, + "step": 36 + }, + { + "epoch": 0.04, + "learning_rate": 0.00010277777777777778, + "loss": 2.1677, + "step": 37 + }, + { + "epoch": 0.04, + "learning_rate": 0.00010555555555555555, + "loss": 2.1255, + "step": 38 + }, + { + "epoch": 0.04, + "learning_rate": 0.00010833333333333334, + "loss": 2.0289, + "step": 39 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001111111111111111, + "loss": 2.0142, + "step": 40 + }, + { + "epoch": 0.04, + "learning_rate": 0.00011388888888888889, + "loss": 1.998, + "step": 41 + }, + { + "epoch": 0.05, + "learning_rate": 0.00011666666666666667, + "loss": 1.9188, + "step": 42 + }, + { + "epoch": 0.05, + "learning_rate": 0.00011944444444444445, + "loss": 1.8715, + "step": 43 + }, + { + "epoch": 0.05, + "learning_rate": 0.00012222222222222221, + "loss": 1.8904, + "step": 44 + }, + { + "epoch": 0.05, + "learning_rate": 0.000125, + "loss": 1.8716, + "step": 45 + }, + { + "epoch": 0.05, + "learning_rate": 0.00012777777777777776, + "loss": 1.7882, + "step": 46 + }, + { + "epoch": 0.05, + "learning_rate": 0.00013055555555555558, + "loss": 1.7679, + "step": 47 + }, + { + "epoch": 0.05, + "learning_rate": 0.00013333333333333334, + "loss": 1.7151, + "step": 48 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001361111111111111, + "loss": 1.6912, + "step": 49 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001388888888888889, + "loss": 1.7754, + "step": 50 + }, + { + "epoch": 0.06, + "learning_rate": 0.00014166666666666668, + "loss": 1.683, + "step": 51 + }, + { + "epoch": 0.06, + "learning_rate": 0.00014444444444444444, + "loss": 1.6863, + "step": 52 + }, + { + "epoch": 0.06, + "learning_rate": 0.00014722222222222223, + "loss": 1.6517, + "step": 53 + }, + { + "epoch": 0.06, + "learning_rate": 0.00015, + "loss": 1.6864, + "step": 54 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001527777777777778, + "loss": 1.6162, + "step": 55 + }, + { + "epoch": 0.06, + "learning_rate": 0.00015555555555555556, + "loss": 1.5718, + "step": 56 + }, + { + "epoch": 0.06, + "learning_rate": 0.00015833333333333332, + "loss": 1.5966, + "step": 57 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001611111111111111, + "loss": 1.5594, + "step": 58 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001638888888888889, + "loss": 1.592, + "step": 59 + }, + { + "epoch": 0.07, + "learning_rate": 0.00016666666666666666, + "loss": 1.5515, + "step": 60 + }, + { + "epoch": 0.07, + "learning_rate": 0.00016944444444444445, + "loss": 1.606, + "step": 61 + }, + { + "epoch": 0.07, + "learning_rate": 0.00017222222222222224, + "loss": 1.5634, + "step": 62 + }, + { + "epoch": 0.07, + "learning_rate": 0.000175, + "loss": 1.5984, + "step": 63 + }, + { + "epoch": 0.07, + "learning_rate": 0.00017777777777777779, + "loss": 1.577, + "step": 64 + }, + { + "epoch": 0.07, + "learning_rate": 0.00018055555555555555, + "loss": 1.58, + "step": 65 + }, + { + "epoch": 0.07, + "learning_rate": 0.00018333333333333334, + "loss": 1.4992, + "step": 66 + }, + { + "epoch": 0.07, + "learning_rate": 0.00018611111111111112, + "loss": 1.5873, + "step": 67 + }, + { + "epoch": 0.07, + "learning_rate": 0.00018888888888888888, + "loss": 1.5352, + "step": 68 + }, + { + "epoch": 0.08, + "learning_rate": 0.00019166666666666667, + "loss": 1.5, + "step": 69 + }, + { + "epoch": 0.08, + "learning_rate": 0.00019444444444444446, + "loss": 1.5277, + "step": 70 + }, + { + "epoch": 0.08, + "learning_rate": 0.00019722222222222222, + "loss": 1.5331, + "step": 71 + }, + { + "epoch": 0.08, + "learning_rate": 0.0002, + "loss": 1.5179, + "step": 72 + }, + { + "epoch": 0.08, + "learning_rate": 0.00020277777777777777, + "loss": 1.4606, + "step": 73 + }, + { + "epoch": 0.08, + "learning_rate": 0.00020555555555555556, + "loss": 1.4401, + "step": 74 + }, + { + "epoch": 0.08, + "learning_rate": 0.00020833333333333335, + "loss": 1.5281, + "step": 75 + }, + { + "epoch": 0.08, + "learning_rate": 0.0002111111111111111, + "loss": 1.4341, + "step": 76 + }, + { + "epoch": 0.08, + "learning_rate": 0.0002138888888888889, + "loss": 1.4533, + "step": 77 + }, + { + "epoch": 0.08, + "learning_rate": 0.00021666666666666668, + "loss": 1.4776, + "step": 78 + }, + { + "epoch": 0.09, + "learning_rate": 0.00021944444444444444, + "loss": 1.5066, + "step": 79 + }, + { + "epoch": 0.09, + "learning_rate": 0.0002222222222222222, + "loss": 1.5293, + "step": 80 + }, + { + "epoch": 0.09, + "learning_rate": 0.00022500000000000002, + "loss": 1.4638, + "step": 81 + }, + { + "epoch": 0.09, + "learning_rate": 0.00022777777777777778, + "loss": 1.4485, + "step": 82 + }, + { + "epoch": 0.09, + "learning_rate": 0.00023055555555555557, + "loss": 1.5267, + "step": 83 + }, + { + "epoch": 0.09, + "learning_rate": 0.00023333333333333333, + "loss": 1.5055, + "step": 84 + }, + { + "epoch": 0.09, + "learning_rate": 0.00023611111111111112, + "loss": 1.4261, + "step": 85 + }, + { + "epoch": 0.09, + "learning_rate": 0.0002388888888888889, + "loss": 1.3983, + "step": 86 + }, + { + "epoch": 0.09, + "learning_rate": 0.00024166666666666667, + "loss": 1.4061, + "step": 87 + }, + { + "epoch": 0.1, + "learning_rate": 0.00024444444444444443, + "loss": 1.4345, + "step": 88 + }, + { + "epoch": 0.1, + "learning_rate": 0.00024722222222222224, + "loss": 1.4546, + "step": 89 + }, + { + "epoch": 0.1, + "learning_rate": 0.00025, + "loss": 1.3898, + "step": 90 + }, + { + "epoch": 0.1, + "learning_rate": 0.00025277777777777777, + "loss": 1.4346, + "step": 91 + }, + { + "epoch": 0.1, + "learning_rate": 0.00025555555555555553, + "loss": 1.4608, + "step": 92 + }, + { + "epoch": 0.1, + "learning_rate": 0.00025833333333333334, + "loss": 1.4902, + "step": 93 + }, + { + "epoch": 0.1, + "learning_rate": 0.00026111111111111116, + "loss": 1.398, + "step": 94 + }, + { + "epoch": 0.1, + "learning_rate": 0.0002638888888888889, + "loss": 1.4108, + "step": 95 + }, + { + "epoch": 0.1, + "learning_rate": 0.0002666666666666667, + "loss": 1.3904, + "step": 96 + }, + { + "epoch": 0.11, + "learning_rate": 0.00026944444444444444, + "loss": 1.3856, + "step": 97 + }, + { + "epoch": 0.11, + "learning_rate": 0.0002722222222222222, + "loss": 1.412, + "step": 98 + }, + { + "epoch": 0.11, + "learning_rate": 0.000275, + "loss": 1.4201, + "step": 99 + }, + { + "epoch": 0.11, + "learning_rate": 0.0002777777777777778, + "loss": 1.3526, + "step": 100 + }, + { + "epoch": 0.11, + "learning_rate": 0.00028055555555555554, + "loss": 1.4473, + "step": 101 + }, + { + "epoch": 0.11, + "learning_rate": 0.00028333333333333335, + "loss": 1.3979, + "step": 102 + }, + { + "epoch": 0.11, + "learning_rate": 0.0002861111111111111, + "loss": 1.3605, + "step": 103 + }, + { + "epoch": 0.11, + "learning_rate": 0.0002888888888888889, + "loss": 1.389, + "step": 104 + }, + { + "epoch": 0.11, + "learning_rate": 0.0002916666666666667, + "loss": 1.3683, + "step": 105 + }, + { + "epoch": 0.12, + "learning_rate": 0.00029444444444444445, + "loss": 1.3876, + "step": 106 + }, + { + "epoch": 0.12, + "learning_rate": 0.0002972222222222222, + "loss": 1.4033, + "step": 107 + }, + { + "epoch": 0.12, + "learning_rate": 0.0003, + "loss": 1.3951, + "step": 108 + }, + { + "epoch": 0.12, + "learning_rate": 0.0003027777777777778, + "loss": 1.3494, + "step": 109 + }, + { + "epoch": 0.12, + "learning_rate": 0.0003055555555555556, + "loss": 1.3452, + "step": 110 + }, + { + "epoch": 0.12, + "learning_rate": 0.00030833333333333337, + "loss": 1.4116, + "step": 111 + }, + { + "epoch": 0.12, + "learning_rate": 0.0003111111111111111, + "loss": 1.4196, + "step": 112 + }, + { + "epoch": 0.12, + "learning_rate": 0.0003138888888888889, + "loss": 1.3441, + "step": 113 + }, + { + "epoch": 0.12, + "learning_rate": 0.00031666666666666665, + "loss": 1.3614, + "step": 114 + }, + { + "epoch": 0.13, + "learning_rate": 0.0003194444444444444, + "loss": 1.3536, + "step": 115 + }, + { + "epoch": 0.13, + "learning_rate": 0.0003222222222222222, + "loss": 1.3626, + "step": 116 + }, + { + "epoch": 0.13, + "learning_rate": 0.00032500000000000004, + "loss": 1.3436, + "step": 117 + }, + { + "epoch": 0.13, + "learning_rate": 0.0003277777777777778, + "loss": 1.3135, + "step": 118 + }, + { + "epoch": 0.13, + "learning_rate": 0.00033055555555555556, + "loss": 1.3916, + "step": 119 + }, + { + "epoch": 0.13, + "learning_rate": 0.0003333333333333333, + "loss": 1.3465, + "step": 120 + }, + { + "epoch": 0.13, + "learning_rate": 0.00033611111111111114, + "loss": 1.3628, + "step": 121 + }, + { + "epoch": 0.13, + "learning_rate": 0.0003388888888888889, + "loss": 1.4076, + "step": 122 + }, + { + "epoch": 0.13, + "learning_rate": 0.00034166666666666666, + "loss": 1.2934, + "step": 123 + }, + { + "epoch": 0.13, + "learning_rate": 0.0003444444444444445, + "loss": 1.3353, + "step": 124 + }, + { + "epoch": 0.14, + "learning_rate": 0.00034722222222222224, + "loss": 1.2837, + "step": 125 + }, + { + "epoch": 0.14, + "learning_rate": 0.00035, + "loss": 1.2845, + "step": 126 + }, + { + "epoch": 0.14, + "learning_rate": 0.0003527777777777778, + "loss": 1.2902, + "step": 127 + }, + { + "epoch": 0.14, + "learning_rate": 0.00035555555555555557, + "loss": 1.3844, + "step": 128 + }, + { + "epoch": 0.14, + "learning_rate": 0.00035833333333333333, + "loss": 1.3711, + "step": 129 + }, + { + "epoch": 0.14, + "learning_rate": 0.0003611111111111111, + "loss": 1.3353, + "step": 130 + }, + { + "epoch": 0.14, + "learning_rate": 0.00036388888888888886, + "loss": 1.3262, + "step": 131 + }, + { + "epoch": 0.14, + "learning_rate": 0.00036666666666666667, + "loss": 1.2868, + "step": 132 + }, + { + "epoch": 0.14, + "learning_rate": 0.0003694444444444445, + "loss": 1.3332, + "step": 133 + }, + { + "epoch": 0.15, + "learning_rate": 0.00037222222222222225, + "loss": 1.2785, + "step": 134 + }, + { + "epoch": 0.15, + "learning_rate": 0.000375, + "loss": 1.3316, + "step": 135 + }, + { + "epoch": 0.15, + "learning_rate": 0.00037777777777777777, + "loss": 1.3902, + "step": 136 + }, + { + "epoch": 0.15, + "learning_rate": 0.00038055555555555553, + "loss": 1.3159, + "step": 137 + }, + { + "epoch": 0.15, + "learning_rate": 0.00038333333333333334, + "loss": 1.3355, + "step": 138 + }, + { + "epoch": 0.15, + "learning_rate": 0.00038611111111111116, + "loss": 1.2676, + "step": 139 + }, + { + "epoch": 0.15, + "learning_rate": 0.0003888888888888889, + "loss": 1.2722, + "step": 140 + }, + { + "epoch": 0.15, + "learning_rate": 0.0003916666666666667, + "loss": 1.3487, + "step": 141 + }, + { + "epoch": 0.15, + "learning_rate": 0.00039444444444444444, + "loss": 1.3746, + "step": 142 + }, + { + "epoch": 0.16, + "learning_rate": 0.0003972222222222222, + "loss": 1.2832, + "step": 143 + }, + { + "epoch": 0.16, + "learning_rate": 0.0004, + "loss": 1.3174, + "step": 144 + }, + { + "epoch": 0.16, + "learning_rate": 0.0004027777777777778, + "loss": 1.3028, + "step": 145 + }, + { + "epoch": 0.16, + "learning_rate": 0.00040555555555555554, + "loss": 1.2666, + "step": 146 + }, + { + "epoch": 0.16, + "learning_rate": 0.00040833333333333336, + "loss": 1.2877, + "step": 147 + }, + { + "epoch": 0.16, + "learning_rate": 0.0004111111111111111, + "loss": 1.2522, + "step": 148 + }, + { + "epoch": 0.16, + "learning_rate": 0.0004138888888888889, + "loss": 1.3421, + "step": 149 + }, + { + "epoch": 0.16, + "learning_rate": 0.0004166666666666667, + "loss": 1.2411, + "step": 150 + }, + { + "epoch": 0.16, + "learning_rate": 0.00041944444444444445, + "loss": 1.2629, + "step": 151 + }, + { + "epoch": 0.17, + "learning_rate": 0.0004222222222222222, + "loss": 1.2925, + "step": 152 + }, + { + "epoch": 0.17, + "learning_rate": 0.000425, + "loss": 1.2424, + "step": 153 + }, + { + "epoch": 0.17, + "learning_rate": 0.0004277777777777778, + "loss": 1.2099, + "step": 154 + }, + { + "epoch": 0.17, + "learning_rate": 0.0004305555555555556, + "loss": 1.232, + "step": 155 + }, + { + "epoch": 0.17, + "learning_rate": 0.00043333333333333337, + "loss": 1.3073, + "step": 156 + }, + { + "epoch": 0.17, + "learning_rate": 0.00043611111111111113, + "loss": 1.2503, + "step": 157 + }, + { + "epoch": 0.17, + "learning_rate": 0.0004388888888888889, + "loss": 1.2967, + "step": 158 + }, + { + "epoch": 0.17, + "learning_rate": 0.00044166666666666665, + "loss": 1.2743, + "step": 159 + }, + { + "epoch": 0.17, + "learning_rate": 0.0004444444444444444, + "loss": 1.2235, + "step": 160 + }, + { + "epoch": 0.18, + "learning_rate": 0.0004472222222222222, + "loss": 1.2514, + "step": 161 + }, + { + "epoch": 0.18, + "learning_rate": 0.00045000000000000004, + "loss": 1.2758, + "step": 162 + }, + { + "epoch": 0.18, + "learning_rate": 0.0004527777777777778, + "loss": 1.2982, + "step": 163 + }, + { + "epoch": 0.18, + "learning_rate": 0.00045555555555555556, + "loss": 1.3149, + "step": 164 + }, + { + "epoch": 0.18, + "learning_rate": 0.0004583333333333333, + "loss": 1.2579, + "step": 165 + }, + { + "epoch": 0.18, + "learning_rate": 0.00046111111111111114, + "loss": 1.2657, + "step": 166 + }, + { + "epoch": 0.18, + "learning_rate": 0.0004638888888888889, + "loss": 1.2372, + "step": 167 + }, + { + "epoch": 0.18, + "learning_rate": 0.00046666666666666666, + "loss": 1.203, + "step": 168 + }, + { + "epoch": 0.18, + "learning_rate": 0.0004694444444444445, + "loss": 1.2448, + "step": 169 + }, + { + "epoch": 0.18, + "learning_rate": 0.00047222222222222224, + "loss": 1.25, + "step": 170 + }, + { + "epoch": 0.19, + "learning_rate": 0.000475, + "loss": 1.2973, + "step": 171 + }, + { + "epoch": 0.19, + "learning_rate": 0.0004777777777777778, + "loss": 1.2044, + "step": 172 + }, + { + "epoch": 0.19, + "learning_rate": 0.0004805555555555556, + "loss": 1.1807, + "step": 173 + }, + { + "epoch": 0.19, + "learning_rate": 0.00048333333333333334, + "loss": 1.2447, + "step": 174 + }, + { + "epoch": 0.19, + "learning_rate": 0.0004861111111111111, + "loss": 1.265, + "step": 175 + }, + { + "epoch": 0.19, + "learning_rate": 0.0004888888888888889, + "loss": 1.2603, + "step": 176 + }, + { + "epoch": 0.19, + "learning_rate": 0.0004916666666666666, + "loss": 1.2455, + "step": 177 + }, + { + "epoch": 0.19, + "learning_rate": 0.0004944444444444445, + "loss": 1.1926, + "step": 178 + }, + { + "epoch": 0.19, + "learning_rate": 0.0004972222222222222, + "loss": 1.2319, + "step": 179 + }, + { + "epoch": 0.2, + "learning_rate": 0.0005, + "loss": 1.2501, + "step": 180 + }, + { + "epoch": 0.2, + "learning_rate": 0.0005, + "loss": 1.2104, + "step": 181 + }, + { + "epoch": 0.2, + "learning_rate": 0.0005, + "loss": 1.1955, + "step": 182 + }, + { + "epoch": 0.2, + "learning_rate": 0.0005, + "loss": 1.1759, + "step": 183 + }, + { + "epoch": 0.2, + "learning_rate": 0.0005, + "loss": 1.2627, + "step": 184 + }, + { + "epoch": 0.2, + "learning_rate": 0.0005, + "loss": 1.1659, + "step": 185 + }, + { + "epoch": 0.2, + "learning_rate": 0.0005, + "loss": 1.2082, + "step": 186 + }, + { + "epoch": 0.2, + "learning_rate": 0.0005, + "loss": 1.2842, + "step": 187 + }, + { + "epoch": 0.2, + "learning_rate": 0.0005, + "loss": 1.1943, + "step": 188 + }, + { + "epoch": 0.21, + "learning_rate": 0.0005, + "loss": 1.2494, + "step": 189 + }, + { + "epoch": 0.21, + "learning_rate": 0.0005, + "loss": 1.1935, + "step": 190 + }, + { + "epoch": 0.21, + "learning_rate": 0.0005, + "loss": 1.1754, + "step": 191 + }, + { + "epoch": 0.21, + "learning_rate": 0.0005, + "loss": 1.2717, + "step": 192 + }, + { + "epoch": 0.21, + "learning_rate": 0.0005, + "loss": 1.2008, + "step": 193 + }, + { + "epoch": 0.21, + "learning_rate": 0.0005, + "loss": 1.2049, + "step": 194 + }, + { + "epoch": 0.21, + "learning_rate": 0.0005, + "loss": 1.182, + "step": 195 + }, + { + "epoch": 0.21, + "learning_rate": 0.0005, + "loss": 1.1936, + "step": 196 + }, + { + "epoch": 0.21, + "learning_rate": 0.0005, + "loss": 1.2024, + "step": 197 + }, + { + "epoch": 0.22, + "learning_rate": 0.0005, + "loss": 1.2534, + "step": 198 + }, + { + "epoch": 0.22, + "learning_rate": 0.0005, + "loss": 1.1986, + "step": 199 + }, + { + "epoch": 0.22, + "learning_rate": 0.0005, + "loss": 1.3031, + "step": 200 + }, + { + "epoch": 0.22, + "learning_rate": 0.0005, + "loss": 1.2503, + "step": 201 + }, + { + "epoch": 0.22, + "learning_rate": 0.0005, + "loss": 1.1916, + "step": 202 + }, + { + "epoch": 0.22, + "learning_rate": 0.0005, + "loss": 1.1704, + "step": 203 + }, + { + "epoch": 0.22, + "learning_rate": 0.0005, + "loss": 1.2644, + "step": 204 + }, + { + "epoch": 0.22, + "learning_rate": 0.0005, + "loss": 1.1649, + "step": 205 + }, + { + "epoch": 0.22, + "learning_rate": 0.0005, + "loss": 1.1897, + "step": 206 + }, + { + "epoch": 0.23, + "learning_rate": 0.0005, + "loss": 1.2378, + "step": 207 + }, + { + "epoch": 0.23, + "learning_rate": 0.0005, + "loss": 1.1909, + "step": 208 + }, + { + "epoch": 0.23, + "learning_rate": 0.0005, + "loss": 1.1687, + "step": 209 + }, + { + "epoch": 0.23, + "learning_rate": 0.0005, + "loss": 1.1617, + "step": 210 + }, + { + "epoch": 0.23, + "learning_rate": 0.0005, + "loss": 1.1845, + "step": 211 + }, + { + "epoch": 0.23, + "learning_rate": 0.0005, + "loss": 1.2246, + "step": 212 + }, + { + "epoch": 0.23, + "learning_rate": 0.0005, + "loss": 1.1967, + "step": 213 + }, + { + "epoch": 0.23, + "learning_rate": 0.0005, + "loss": 1.1896, + "step": 214 + }, + { + "epoch": 0.23, + "learning_rate": 0.0005, + "loss": 1.2793, + "step": 215 + }, + { + "epoch": 0.24, + "learning_rate": 0.0005, + "loss": 1.2362, + "step": 216 + }, + { + "epoch": 0.24, + "learning_rate": 0.0005, + "loss": 1.1963, + "step": 217 + }, + { + "epoch": 0.24, + "learning_rate": 0.0005, + "loss": 1.1692, + "step": 218 + }, + { + "epoch": 0.24, + "learning_rate": 0.0005, + "loss": 1.215, + "step": 219 + }, + { + "epoch": 0.24, + "learning_rate": 0.0005, + "loss": 1.1741, + "step": 220 + }, + { + "epoch": 0.24, + "learning_rate": 0.0005, + "loss": 1.2078, + "step": 221 + }, + { + "epoch": 0.24, + "learning_rate": 0.0005, + "loss": 1.1777, + "step": 222 + }, + { + "epoch": 0.24, + "learning_rate": 0.0005, + "loss": 1.2123, + "step": 223 + }, + { + "epoch": 0.24, + "learning_rate": 0.0005, + "loss": 1.1975, + "step": 224 + }, + { + "epoch": 0.24, + "learning_rate": 0.0005, + "loss": 1.2305, + "step": 225 + }, + { + "epoch": 0.25, + "learning_rate": 0.0005, + "loss": 1.1652, + "step": 226 + }, + { + "epoch": 0.25, + "learning_rate": 0.0005, + "loss": 1.1781, + "step": 227 + }, + { + "epoch": 0.25, + "learning_rate": 0.0005, + "loss": 1.1968, + "step": 228 + }, + { + "epoch": 0.25, + "learning_rate": 0.0005, + "loss": 1.1425, + "step": 229 + }, + { + "epoch": 0.25, + "learning_rate": 0.0005, + "loss": 1.1933, + "step": 230 + }, + { + "epoch": 0.25, + "learning_rate": 0.0005, + "loss": 1.151, + "step": 231 + }, + { + "epoch": 0.25, + "learning_rate": 0.0005, + "loss": 1.1923, + "step": 232 + }, + { + "epoch": 0.25, + "learning_rate": 0.0005, + "loss": 1.2289, + "step": 233 + }, + { + "epoch": 0.25, + "learning_rate": 0.0005, + "loss": 1.1526, + "step": 234 + }, + { + "epoch": 0.26, + "learning_rate": 0.0005, + "loss": 1.2487, + "step": 235 + }, + { + "epoch": 0.26, + "learning_rate": 0.0005, + "loss": 1.1658, + "step": 236 + }, + { + "epoch": 0.26, + "learning_rate": 0.0005, + "loss": 1.1752, + "step": 237 + }, + { + "epoch": 0.26, + "learning_rate": 0.0005, + "loss": 1.2331, + "step": 238 + }, + { + "epoch": 0.26, + "learning_rate": 0.0005, + "loss": 1.2081, + "step": 239 + }, + { + "epoch": 0.26, + "learning_rate": 0.0005, + "loss": 1.1688, + "step": 240 + }, + { + "epoch": 0.26, + "learning_rate": 0.0005, + "loss": 1.1753, + "step": 241 + }, + { + "epoch": 0.26, + "learning_rate": 0.0005, + "loss": 1.1844, + "step": 242 + }, + { + "epoch": 0.26, + "learning_rate": 0.0005, + "loss": 1.233, + "step": 243 + }, + { + "epoch": 0.27, + "learning_rate": 0.0005, + "loss": 1.1607, + "step": 244 + }, + { + "epoch": 0.27, + "learning_rate": 0.0005, + "loss": 1.1212, + "step": 245 + }, + { + "epoch": 0.27, + "learning_rate": 0.0005, + "loss": 1.1657, + "step": 246 + }, + { + "epoch": 0.27, + "learning_rate": 0.0005, + "loss": 1.1642, + "step": 247 + }, + { + "epoch": 0.27, + "learning_rate": 0.0005, + "loss": 1.2154, + "step": 248 + }, + { + "epoch": 0.27, + "learning_rate": 0.0005, + "loss": 1.1691, + "step": 249 + }, + { + "epoch": 0.27, + "learning_rate": 0.0005, + "loss": 1.1813, + "step": 250 + }, + { + "epoch": 0.27, + "learning_rate": 0.0005, + "loss": 1.1523, + "step": 251 + }, + { + "epoch": 0.27, + "learning_rate": 0.0005, + "loss": 1.1322, + "step": 252 + }, + { + "epoch": 0.28, + "learning_rate": 0.0005, + "loss": 1.2245, + "step": 253 + }, + { + "epoch": 0.28, + "learning_rate": 0.0005, + "loss": 1.1752, + "step": 254 + }, + { + "epoch": 0.28, + "learning_rate": 0.0005, + "loss": 1.1596, + "step": 255 + }, + { + "epoch": 0.28, + "learning_rate": 0.0005, + "loss": 1.1393, + "step": 256 + }, + { + "epoch": 0.28, + "learning_rate": 0.0005, + "loss": 1.1568, + "step": 257 + }, + { + "epoch": 0.28, + "learning_rate": 0.0005, + "loss": 1.188, + "step": 258 + }, + { + "epoch": 0.28, + "learning_rate": 0.0005, + "loss": 1.152, + "step": 259 + }, + { + "epoch": 0.28, + "learning_rate": 0.0005, + "loss": 1.1747, + "step": 260 + }, + { + "epoch": 0.28, + "learning_rate": 0.0005, + "loss": 1.1492, + "step": 261 + }, + { + "epoch": 0.29, + "learning_rate": 0.0005, + "loss": 1.1867, + "step": 262 + }, + { + "epoch": 0.29, + "learning_rate": 0.0005, + "loss": 1.1415, + "step": 263 + }, + { + "epoch": 0.29, + "learning_rate": 0.0005, + "loss": 1.0918, + "step": 264 + }, + { + "epoch": 0.29, + "learning_rate": 0.0005, + "loss": 1.1829, + "step": 265 + }, + { + "epoch": 0.29, + "learning_rate": 0.0005, + "loss": 1.1251, + "step": 266 + }, + { + "epoch": 0.29, + "learning_rate": 0.0005, + "loss": 1.1948, + "step": 267 + }, + { + "epoch": 0.29, + "learning_rate": 0.0005, + "loss": 1.0871, + "step": 268 + }, + { + "epoch": 0.29, + "learning_rate": 0.0005, + "loss": 1.1583, + "step": 269 + }, + { + "epoch": 0.29, + "learning_rate": 0.0005, + "loss": 1.1035, + "step": 270 + }, + { + "epoch": 0.29, + "learning_rate": 0.0005, + "loss": 1.2108, + "step": 271 + }, + { + "epoch": 0.3, + "learning_rate": 0.0005, + "loss": 1.1617, + "step": 272 + }, + { + "epoch": 0.3, + "learning_rate": 0.0005, + "loss": 1.1415, + "step": 273 + }, + { + "epoch": 0.3, + "learning_rate": 0.0005, + "loss": 1.1835, + "step": 274 + }, + { + "epoch": 0.3, + "learning_rate": 0.0005, + "loss": 1.1669, + "step": 275 + }, + { + "epoch": 0.3, + "learning_rate": 0.0005, + "loss": 1.1589, + "step": 276 + }, + { + "epoch": 0.3, + "learning_rate": 0.0005, + "loss": 1.1551, + "step": 277 + }, + { + "epoch": 0.3, + "learning_rate": 0.0005, + "loss": 1.1896, + "step": 278 + }, + { + "epoch": 0.3, + "learning_rate": 0.0005, + "loss": 1.1635, + "step": 279 + }, + { + "epoch": 0.3, + "learning_rate": 0.0005, + "loss": 1.1318, + "step": 280 + }, + { + "epoch": 0.31, + "learning_rate": 0.0005, + "loss": 1.1901, + "step": 281 + }, + { + "epoch": 0.31, + "learning_rate": 0.0005, + "loss": 1.1671, + "step": 282 + }, + { + "epoch": 0.31, + "learning_rate": 0.0005, + "loss": 1.1711, + "step": 283 + }, + { + "epoch": 0.31, + "learning_rate": 0.0005, + "loss": 1.121, + "step": 284 + }, + { + "epoch": 0.31, + "learning_rate": 0.0005, + "loss": 1.1899, + "step": 285 + }, + { + "epoch": 0.31, + "learning_rate": 0.0005, + "loss": 1.1769, + "step": 286 + }, + { + "epoch": 0.31, + "learning_rate": 0.0005, + "loss": 1.1859, + "step": 287 + }, + { + "epoch": 0.31, + "learning_rate": 0.0005, + "loss": 1.1547, + "step": 288 + }, + { + "epoch": 0.31, + "learning_rate": 0.0005, + "loss": 1.0976, + "step": 289 + }, + { + "epoch": 0.32, + "learning_rate": 0.0005, + "loss": 1.1912, + "step": 290 + }, + { + "epoch": 0.32, + "learning_rate": 0.0005, + "loss": 1.2091, + "step": 291 + }, + { + "epoch": 0.32, + "learning_rate": 0.0005, + "loss": 1.1092, + "step": 292 + }, + { + "epoch": 0.32, + "learning_rate": 0.0005, + "loss": 1.1147, + "step": 293 + }, + { + "epoch": 0.32, + "learning_rate": 0.0005, + "loss": 1.1473, + "step": 294 + }, + { + "epoch": 0.32, + "learning_rate": 0.0005, + "loss": 1.1025, + "step": 295 + }, + { + "epoch": 0.32, + "learning_rate": 0.0005, + "loss": 1.1653, + "step": 296 + }, + { + "epoch": 0.32, + "learning_rate": 0.0005, + "loss": 1.1267, + "step": 297 + }, + { + "epoch": 0.32, + "learning_rate": 0.0005, + "loss": 1.1704, + "step": 298 + }, + { + "epoch": 0.33, + "learning_rate": 0.0005, + "loss": 1.1654, + "step": 299 + }, + { + "epoch": 0.33, + "learning_rate": 0.0005, + "loss": 1.17, + "step": 300 + }, + { + "epoch": 0.33, + "learning_rate": 0.0005, + "loss": 1.1459, + "step": 301 + }, + { + "epoch": 0.33, + "learning_rate": 0.0005, + "loss": 1.154, + "step": 302 + }, + { + "epoch": 0.33, + "learning_rate": 0.0005, + "loss": 1.1272, + "step": 303 + }, + { + "epoch": 0.33, + "learning_rate": 0.0005, + "loss": 1.1724, + "step": 304 + }, + { + "epoch": 0.33, + "learning_rate": 0.0005, + "loss": 1.2047, + "step": 305 + }, + { + "epoch": 0.33, + "learning_rate": 0.0005, + "loss": 1.1838, + "step": 306 + }, + { + "epoch": 0.33, + "learning_rate": 0.0005, + "loss": 1.1326, + "step": 307 + }, + { + "epoch": 0.34, + "learning_rate": 0.0005, + "loss": 1.1063, + "step": 308 + }, + { + "epoch": 0.34, + "learning_rate": 0.0005, + "loss": 1.1382, + "step": 309 + }, + { + "epoch": 0.34, + "learning_rate": 0.0005, + "loss": 1.2094, + "step": 310 + }, + { + "epoch": 0.34, + "learning_rate": 0.0005, + "loss": 1.1924, + "step": 311 + }, + { + "epoch": 0.34, + "learning_rate": 0.0005, + "loss": 1.1513, + "step": 312 + }, + { + "epoch": 0.34, + "learning_rate": 0.0005, + "loss": 1.1471, + "step": 313 + }, + { + "epoch": 0.34, + "learning_rate": 0.0005, + "loss": 1.1268, + "step": 314 + }, + { + "epoch": 0.34, + "learning_rate": 0.0005, + "loss": 1.1566, + "step": 315 + }, + { + "epoch": 0.34, + "learning_rate": 0.0005, + "loss": 1.0798, + "step": 316 + }, + { + "epoch": 0.34, + "learning_rate": 0.0005, + "loss": 1.267, + "step": 317 + }, + { + "epoch": 0.35, + "learning_rate": 0.0005, + "loss": 1.1165, + "step": 318 + }, + { + "epoch": 0.35, + "learning_rate": 0.0005, + "loss": 1.0973, + "step": 319 + }, + { + "epoch": 0.35, + "learning_rate": 0.0005, + "loss": 1.1077, + "step": 320 + }, + { + "epoch": 0.35, + "learning_rate": 0.0005, + "loss": 1.1538, + "step": 321 + }, + { + "epoch": 0.35, + "learning_rate": 0.0005, + "loss": 1.1111, + "step": 322 + }, + { + "epoch": 0.35, + "learning_rate": 0.0005, + "loss": 1.1497, + "step": 323 + }, + { + "epoch": 0.35, + "learning_rate": 0.0005, + "loss": 1.1579, + "step": 324 + }, + { + "epoch": 0.35, + "learning_rate": 0.0005, + "loss": 1.1407, + "step": 325 + }, + { + "epoch": 0.35, + "learning_rate": 0.0005, + "loss": 1.1359, + "step": 326 + }, + { + "epoch": 0.36, + "learning_rate": 0.0005, + "loss": 1.1745, + "step": 327 + }, + { + "epoch": 0.36, + "learning_rate": 0.0005, + "loss": 1.167, + "step": 328 + }, + { + "epoch": 0.36, + "learning_rate": 0.0005, + "loss": 1.1816, + "step": 329 + }, + { + "epoch": 0.36, + "learning_rate": 0.0005, + "loss": 1.169, + "step": 330 + }, + { + "epoch": 0.36, + "learning_rate": 0.0005, + "loss": 1.119, + "step": 331 + }, + { + "epoch": 0.36, + "learning_rate": 0.0005, + "loss": 1.1806, + "step": 332 + }, + { + "epoch": 0.36, + "learning_rate": 0.0005, + "loss": 1.1916, + "step": 333 + }, + { + "epoch": 0.36, + "learning_rate": 0.0005, + "loss": 1.1489, + "step": 334 + }, + { + "epoch": 0.36, + "learning_rate": 0.0005, + "loss": 1.133, + "step": 335 + }, + { + "epoch": 0.37, + "learning_rate": 0.0005, + "loss": 1.1741, + "step": 336 + }, + { + "epoch": 0.37, + "learning_rate": 0.0005, + "loss": 1.1132, + "step": 337 + }, + { + "epoch": 0.37, + "learning_rate": 0.0005, + "loss": 1.1262, + "step": 338 + }, + { + "epoch": 0.37, + "learning_rate": 0.0005, + "loss": 1.1072, + "step": 339 + }, + { + "epoch": 0.37, + "learning_rate": 0.0005, + "loss": 1.124, + "step": 340 + }, + { + "epoch": 0.37, + "learning_rate": 0.0005, + "loss": 1.1135, + "step": 341 + }, + { + "epoch": 0.37, + "learning_rate": 0.0005, + "loss": 1.1375, + "step": 342 + }, + { + "epoch": 0.37, + "learning_rate": 0.0005, + "loss": 1.1816, + "step": 343 + }, + { + "epoch": 0.37, + "learning_rate": 0.0005, + "loss": 1.1213, + "step": 344 + }, + { + "epoch": 0.38, + "learning_rate": 0.0005, + "loss": 1.1505, + "step": 345 + }, + { + "epoch": 0.38, + "learning_rate": 0.0005, + "loss": 1.2095, + "step": 346 + }, + { + "epoch": 0.38, + "learning_rate": 0.0005, + "loss": 1.1606, + "step": 347 + }, + { + "epoch": 0.38, + "learning_rate": 0.0005, + "loss": 1.1337, + "step": 348 + }, + { + "epoch": 0.38, + "learning_rate": 0.0005, + "loss": 1.1569, + "step": 349 + }, + { + "epoch": 0.38, + "learning_rate": 0.0005, + "loss": 1.159, + "step": 350 + }, + { + "epoch": 0.38, + "learning_rate": 0.0005, + "loss": 1.109, + "step": 351 + }, + { + "epoch": 0.38, + "learning_rate": 0.0005, + "loss": 1.1253, + "step": 352 + }, + { + "epoch": 0.38, + "learning_rate": 0.0005, + "loss": 1.1293, + "step": 353 + }, + { + "epoch": 0.39, + "learning_rate": 0.0005, + "loss": 1.1065, + "step": 354 + }, + { + "epoch": 0.39, + "learning_rate": 0.0005, + "loss": 1.1155, + "step": 355 + }, + { + "epoch": 0.39, + "learning_rate": 0.0005, + "loss": 1.1258, + "step": 356 + }, + { + "epoch": 0.39, + "learning_rate": 0.0005, + "loss": 1.1052, + "step": 357 + }, + { + "epoch": 0.39, + "learning_rate": 0.0005, + "loss": 1.1212, + "step": 358 + }, + { + "epoch": 0.39, + "learning_rate": 0.0005, + "loss": 1.095, + "step": 359 + }, + { + "epoch": 0.39, + "learning_rate": 0.0005, + "loss": 1.1834, + "step": 360 + }, + { + "epoch": 0.39, + "learning_rate": 0.0005, + "loss": 1.1111, + "step": 361 + }, + { + "epoch": 0.39, + "learning_rate": 0.0005, + "loss": 1.1592, + "step": 362 + }, + { + "epoch": 0.39, + "learning_rate": 0.0005, + "loss": 1.1187, + "step": 363 + }, + { + "epoch": 0.4, + "learning_rate": 0.0005, + "loss": 1.1215, + "step": 364 + }, + { + "epoch": 0.4, + "learning_rate": 0.0005, + "loss": 1.1184, + "step": 365 + }, + { + "epoch": 0.4, + "learning_rate": 0.0005, + "loss": 1.0882, + "step": 366 + }, + { + "epoch": 0.4, + "learning_rate": 0.0005, + "loss": 1.1468, + "step": 367 + }, + { + "epoch": 0.4, + "learning_rate": 0.0005, + "loss": 1.149, + "step": 368 + }, + { + "epoch": 0.4, + "learning_rate": 0.0005, + "loss": 1.1485, + "step": 369 + }, + { + "epoch": 0.4, + "learning_rate": 0.0005, + "loss": 1.1312, + "step": 370 + }, + { + "epoch": 0.4, + "learning_rate": 0.0005, + "loss": 1.0817, + "step": 371 + }, + { + "epoch": 0.4, + "learning_rate": 0.0005, + "loss": 1.1428, + "step": 372 + }, + { + "epoch": 0.41, + "learning_rate": 0.0005, + "loss": 1.0791, + "step": 373 + }, + { + "epoch": 0.41, + "learning_rate": 0.0005, + "loss": 1.1449, + "step": 374 + }, + { + "epoch": 0.41, + "learning_rate": 0.0005, + "loss": 1.1281, + "step": 375 + }, + { + "epoch": 0.41, + "learning_rate": 0.0005, + "loss": 1.1784, + "step": 376 + }, + { + "epoch": 0.41, + "learning_rate": 0.0005, + "loss": 1.1194, + "step": 377 + }, + { + "epoch": 0.41, + "learning_rate": 0.0005, + "loss": 1.1417, + "step": 378 + }, + { + "epoch": 0.41, + "learning_rate": 0.0005, + "loss": 1.1331, + "step": 379 + }, + { + "epoch": 0.41, + "learning_rate": 0.0005, + "loss": 1.1249, + "step": 380 + }, + { + "epoch": 0.41, + "learning_rate": 0.0005, + "loss": 1.1208, + "step": 381 + }, + { + "epoch": 0.42, + "learning_rate": 0.0005, + "loss": 1.1591, + "step": 382 + }, + { + "epoch": 0.42, + "learning_rate": 0.0005, + "loss": 1.1562, + "step": 383 + }, + { + "epoch": 0.42, + "learning_rate": 0.0005, + "loss": 1.1948, + "step": 384 + }, + { + "epoch": 0.42, + "learning_rate": 0.0005, + "loss": 1.1647, + "step": 385 + }, + { + "epoch": 0.42, + "learning_rate": 0.0005, + "loss": 1.0564, + "step": 386 + }, + { + "epoch": 0.42, + "learning_rate": 0.0005, + "loss": 1.1051, + "step": 387 + }, + { + "epoch": 0.42, + "learning_rate": 0.0005, + "loss": 1.11, + "step": 388 + }, + { + "epoch": 0.42, + "learning_rate": 0.0005, + "loss": 1.0871, + "step": 389 + }, + { + "epoch": 0.42, + "learning_rate": 0.0005, + "loss": 1.0904, + "step": 390 + }, + { + "epoch": 0.43, + "learning_rate": 0.0005, + "loss": 1.1267, + "step": 391 + }, + { + "epoch": 0.43, + "learning_rate": 0.0005, + "loss": 1.1594, + "step": 392 + }, + { + "epoch": 0.43, + "learning_rate": 0.0005, + "loss": 1.1525, + "step": 393 + }, + { + "epoch": 0.43, + "learning_rate": 0.0005, + "loss": 1.0888, + "step": 394 + }, + { + "epoch": 0.43, + "learning_rate": 0.0005, + "loss": 1.1201, + "step": 395 + }, + { + "epoch": 0.43, + "learning_rate": 0.0005, + "loss": 1.1309, + "step": 396 + }, + { + "epoch": 0.43, + "learning_rate": 0.0005, + "loss": 1.1618, + "step": 397 + }, + { + "epoch": 0.43, + "learning_rate": 0.0005, + "loss": 1.1327, + "step": 398 + }, + { + "epoch": 0.43, + "learning_rate": 0.0005, + "loss": 1.1383, + "step": 399 + }, + { + "epoch": 0.44, + "learning_rate": 0.0005, + "loss": 1.1226, + "step": 400 + }, + { + "epoch": 0.44, + "learning_rate": 0.0005, + "loss": 1.1208, + "step": 401 + }, + { + "epoch": 0.44, + "learning_rate": 0.0005, + "loss": 1.1492, + "step": 402 + }, + { + "epoch": 0.44, + "learning_rate": 0.0005, + "loss": 1.1441, + "step": 403 + }, + { + "epoch": 0.44, + "learning_rate": 0.0005, + "loss": 1.1742, + "step": 404 + }, + { + "epoch": 0.44, + "learning_rate": 0.0005, + "loss": 1.1154, + "step": 405 + }, + { + "epoch": 0.44, + "learning_rate": 0.0005, + "loss": 1.1704, + "step": 406 + }, + { + "epoch": 0.44, + "learning_rate": 0.0005, + "loss": 1.0913, + "step": 407 + }, + { + "epoch": 0.44, + "learning_rate": 0.0005, + "loss": 1.1144, + "step": 408 + }, + { + "epoch": 0.45, + "learning_rate": 0.0005, + "loss": 1.1461, + "step": 409 + }, + { + "epoch": 0.45, + "learning_rate": 0.0005, + "loss": 1.186, + "step": 410 + }, + { + "epoch": 0.45, + "learning_rate": 0.0005, + "loss": 1.1279, + "step": 411 + }, + { + "epoch": 0.45, + "learning_rate": 0.0005, + "loss": 1.1027, + "step": 412 + }, + { + "epoch": 0.45, + "learning_rate": 0.0005, + "loss": 1.0754, + "step": 413 + }, + { + "epoch": 0.45, + "learning_rate": 0.0005, + "loss": 1.1038, + "step": 414 + }, + { + "epoch": 0.45, + "learning_rate": 0.0005, + "loss": 1.1766, + "step": 415 + }, + { + "epoch": 0.45, + "learning_rate": 0.0005, + "loss": 1.1195, + "step": 416 + }, + { + "epoch": 0.45, + "learning_rate": 0.0005, + "loss": 1.136, + "step": 417 + }, + { + "epoch": 0.45, + "learning_rate": 0.0005, + "loss": 1.148, + "step": 418 + }, + { + "epoch": 0.46, + "learning_rate": 0.0005, + "loss": 1.1444, + "step": 419 + }, + { + "epoch": 0.46, + "learning_rate": 0.0005, + "loss": 1.197, + "step": 420 + }, + { + "epoch": 0.46, + "learning_rate": 0.0005, + "loss": 1.1144, + "step": 421 + }, + { + "epoch": 0.46, + "learning_rate": 0.0005, + "loss": 1.141, + "step": 422 + }, + { + "epoch": 0.46, + "learning_rate": 0.0005, + "loss": 1.1381, + "step": 423 + }, + { + "epoch": 0.46, + "learning_rate": 0.0005, + "loss": 1.191, + "step": 424 + }, + { + "epoch": 0.46, + "learning_rate": 0.0005, + "loss": 1.0779, + "step": 425 + }, + { + "epoch": 0.46, + "learning_rate": 0.0005, + "loss": 1.1022, + "step": 426 + }, + { + "epoch": 0.46, + "learning_rate": 0.0005, + "loss": 1.0886, + "step": 427 + }, + { + "epoch": 0.47, + "learning_rate": 0.0005, + "loss": 1.1454, + "step": 428 + }, + { + "epoch": 0.47, + "learning_rate": 0.0005, + "loss": 1.1257, + "step": 429 + }, + { + "epoch": 0.47, + "learning_rate": 0.0005, + "loss": 1.0935, + "step": 430 + }, + { + "epoch": 0.47, + "learning_rate": 0.0005, + "loss": 1.0932, + "step": 431 + }, + { + "epoch": 0.47, + "learning_rate": 0.0005, + "loss": 1.1824, + "step": 432 + }, + { + "epoch": 0.47, + "learning_rate": 0.0005, + "loss": 1.1599, + "step": 433 + }, + { + "epoch": 0.47, + "learning_rate": 0.0005, + "loss": 1.1213, + "step": 434 + }, + { + "epoch": 0.47, + "learning_rate": 0.0005, + "loss": 1.1425, + "step": 435 + }, + { + "epoch": 0.47, + "learning_rate": 0.0005, + "loss": 1.094, + "step": 436 + }, + { + "epoch": 0.48, + "learning_rate": 0.0005, + "loss": 1.1431, + "step": 437 + }, + { + "epoch": 0.48, + "learning_rate": 0.0005, + "loss": 1.1075, + "step": 438 + }, + { + "epoch": 0.48, + "learning_rate": 0.0005, + "loss": 1.1029, + "step": 439 + }, + { + "epoch": 0.48, + "learning_rate": 0.0005, + "loss": 1.0991, + "step": 440 + }, + { + "epoch": 0.48, + "learning_rate": 0.0005, + "loss": 1.107, + "step": 441 + }, + { + "epoch": 0.48, + "learning_rate": 0.0005, + "loss": 1.1108, + "step": 442 + }, + { + "epoch": 0.48, + "learning_rate": 0.0005, + "loss": 1.1109, + "step": 443 + }, + { + "epoch": 0.48, + "learning_rate": 0.0005, + "loss": 1.1001, + "step": 444 + }, + { + "epoch": 0.48, + "learning_rate": 0.0005, + "loss": 1.1122, + "step": 445 + }, + { + "epoch": 0.49, + "learning_rate": 0.0005, + "loss": 1.1647, + "step": 446 + }, + { + "epoch": 0.49, + "learning_rate": 0.0005, + "loss": 1.1571, + "step": 447 + }, + { + "epoch": 0.49, + "learning_rate": 0.0005, + "loss": 1.1203, + "step": 448 + }, + { + "epoch": 0.49, + "learning_rate": 0.0005, + "loss": 1.1374, + "step": 449 + }, + { + "epoch": 0.49, + "learning_rate": 0.0005, + "loss": 1.1189, + "step": 450 + }, + { + "epoch": 0.49, + "learning_rate": 0.0005, + "loss": 1.1531, + "step": 451 + }, + { + "epoch": 0.49, + "learning_rate": 0.0005, + "loss": 1.1362, + "step": 452 + }, + { + "epoch": 0.49, + "learning_rate": 0.0005, + "loss": 1.1746, + "step": 453 + }, + { + "epoch": 0.49, + "learning_rate": 0.0005, + "loss": 1.149, + "step": 454 + }, + { + "epoch": 0.5, + "learning_rate": 0.0005, + "loss": 1.1452, + "step": 455 + }, + { + "epoch": 0.5, + "learning_rate": 0.0005, + "loss": 1.1029, + "step": 456 + }, + { + "epoch": 0.5, + "learning_rate": 0.0005, + "loss": 1.1254, + "step": 457 + }, + { + "epoch": 0.5, + "learning_rate": 0.0005, + "loss": 1.0814, + "step": 458 + }, + { + "epoch": 0.5, + "learning_rate": 0.0005, + "loss": 1.1769, + "step": 459 + }, + { + "epoch": 0.5, + "learning_rate": 0.0005, + "loss": 1.1446, + "step": 460 + }, + { + "epoch": 0.5, + "learning_rate": 0.0005, + "loss": 1.1435, + "step": 461 + }, + { + "epoch": 0.5, + "learning_rate": 0.0005, + "loss": 1.1249, + "step": 462 + }, + { + "epoch": 0.5, + "learning_rate": 0.0005, + "loss": 1.0767, + "step": 463 + }, + { + "epoch": 0.5, + "learning_rate": 0.0005, + "loss": 1.0756, + "step": 464 + }, + { + "epoch": 0.51, + "learning_rate": 0.0005, + "loss": 1.1044, + "step": 465 + }, + { + "epoch": 0.51, + "learning_rate": 0.0005, + "loss": 1.1442, + "step": 466 + }, + { + "epoch": 0.51, + "learning_rate": 0.0005, + "loss": 1.1533, + "step": 467 + }, + { + "epoch": 0.51, + "learning_rate": 0.0005, + "loss": 1.1731, + "step": 468 + }, + { + "epoch": 0.51, + "learning_rate": 0.0005, + "loss": 1.1051, + "step": 469 + }, + { + "epoch": 0.51, + "learning_rate": 0.0005, + "loss": 1.1828, + "step": 470 + }, + { + "epoch": 0.51, + "learning_rate": 0.0005, + "loss": 1.1326, + "step": 471 + }, + { + "epoch": 0.51, + "learning_rate": 0.0005, + "loss": 1.1079, + "step": 472 + }, + { + "epoch": 0.51, + "learning_rate": 0.0005, + "loss": 1.1553, + "step": 473 + }, + { + "epoch": 0.52, + "learning_rate": 0.0005, + "loss": 1.1859, + "step": 474 + }, + { + "epoch": 0.52, + "learning_rate": 0.0005, + "loss": 1.1411, + "step": 475 + }, + { + "epoch": 0.52, + "learning_rate": 0.0005, + "loss": 1.0923, + "step": 476 + }, + { + "epoch": 0.52, + "learning_rate": 0.0005, + "loss": 1.1239, + "step": 477 + }, + { + "epoch": 0.52, + "learning_rate": 0.0005, + "loss": 1.18, + "step": 478 + }, + { + "epoch": 0.52, + "learning_rate": 0.0005, + "loss": 1.1454, + "step": 479 + }, + { + "epoch": 0.52, + "learning_rate": 0.0005, + "loss": 1.1222, + "step": 480 + }, + { + "epoch": 0.52, + "learning_rate": 0.0005, + "loss": 1.1965, + "step": 481 + }, + { + "epoch": 0.52, + "learning_rate": 0.0005, + "loss": 1.1057, + "step": 482 + }, + { + "epoch": 0.53, + "learning_rate": 0.0005, + "loss": 1.128, + "step": 483 + }, + { + "epoch": 0.53, + "learning_rate": 0.0005, + "loss": 1.13, + "step": 484 + }, + { + "epoch": 0.53, + "learning_rate": 0.0005, + "loss": 1.129, + "step": 485 + }, + { + "epoch": 0.53, + "learning_rate": 0.0005, + "loss": 1.0847, + "step": 486 + }, + { + "epoch": 0.53, + "learning_rate": 0.0005, + "loss": 1.1389, + "step": 487 + }, + { + "epoch": 0.53, + "learning_rate": 0.0005, + "loss": 1.1743, + "step": 488 + }, + { + "epoch": 0.53, + "learning_rate": 0.0005, + "loss": 1.1498, + "step": 489 + }, + { + "epoch": 0.53, + "learning_rate": 0.0005, + "loss": 1.1089, + "step": 490 + }, + { + "epoch": 0.53, + "learning_rate": 0.0005, + "loss": 1.1288, + "step": 491 + }, + { + "epoch": 0.54, + "learning_rate": 0.0005, + "loss": 1.1073, + "step": 492 + }, + { + "epoch": 0.54, + "learning_rate": 0.0005, + "loss": 1.1023, + "step": 493 + }, + { + "epoch": 0.54, + "learning_rate": 0.0005, + "loss": 1.1778, + "step": 494 + }, + { + "epoch": 0.54, + "learning_rate": 0.0005, + "loss": 1.1236, + "step": 495 + }, + { + "epoch": 0.54, + "learning_rate": 0.0005, + "loss": 1.129, + "step": 496 + }, + { + "epoch": 0.54, + "learning_rate": 0.0005, + "loss": 1.1088, + "step": 497 + }, + { + "epoch": 0.54, + "learning_rate": 0.0005, + "loss": 1.1641, + "step": 498 + }, + { + "epoch": 0.54, + "learning_rate": 0.0005, + "loss": 1.0958, + "step": 499 + }, + { + "epoch": 0.54, + "learning_rate": 0.0005, + "loss": 1.0925, + "step": 500 + }, + { + "epoch": 0.55, + "learning_rate": 0.0005, + "loss": 1.1208, + "step": 501 + }, + { + "epoch": 0.55, + "learning_rate": 0.0005, + "loss": 1.1541, + "step": 502 + }, + { + "epoch": 0.55, + "learning_rate": 0.0005, + "loss": 1.0887, + "step": 503 + }, + { + "epoch": 0.55, + "learning_rate": 0.0005, + "loss": 1.1331, + "step": 504 + }, + { + "epoch": 0.55, + "learning_rate": 0.0005, + "loss": 1.1128, + "step": 505 + }, + { + "epoch": 0.55, + "learning_rate": 0.0005, + "loss": 1.0945, + "step": 506 + }, + { + "epoch": 0.55, + "learning_rate": 0.0005, + "loss": 1.1026, + "step": 507 + }, + { + "epoch": 0.55, + "learning_rate": 0.0005, + "loss": 1.1539, + "step": 508 + }, + { + "epoch": 0.55, + "learning_rate": 0.0005, + "loss": 1.1286, + "step": 509 + }, + { + "epoch": 0.55, + "learning_rate": 0.0005, + "loss": 1.0453, + "step": 510 + }, + { + "epoch": 0.56, + "learning_rate": 0.0005, + "loss": 1.1376, + "step": 511 + }, + { + "epoch": 0.56, + "learning_rate": 0.0005, + "loss": 1.1523, + "step": 512 + }, + { + "epoch": 0.56, + "learning_rate": 0.0005, + "loss": 1.0747, + "step": 513 + }, + { + "epoch": 0.56, + "learning_rate": 0.0005, + "loss": 1.1524, + "step": 514 + }, + { + "epoch": 0.56, + "learning_rate": 0.0005, + "loss": 1.1498, + "step": 515 + }, + { + "epoch": 0.56, + "learning_rate": 0.0005, + "loss": 1.1406, + "step": 516 + }, + { + "epoch": 0.56, + "learning_rate": 0.0005, + "loss": 1.1667, + "step": 517 + }, + { + "epoch": 0.56, + "learning_rate": 0.0005, + "loss": 1.1435, + "step": 518 + }, + { + "epoch": 0.56, + "learning_rate": 0.0005, + "loss": 1.1327, + "step": 519 + }, + { + "epoch": 0.57, + "learning_rate": 0.0005, + "loss": 1.1433, + "step": 520 + }, + { + "epoch": 0.57, + "learning_rate": 0.0005, + "loss": 1.1048, + "step": 521 + }, + { + "epoch": 0.57, + "learning_rate": 0.0005, + "loss": 1.1297, + "step": 522 + }, + { + "epoch": 0.57, + "learning_rate": 0.0005, + "loss": 1.1026, + "step": 523 + }, + { + "epoch": 0.57, + "learning_rate": 0.0005, + "loss": 1.1478, + "step": 524 + }, + { + "epoch": 0.57, + "learning_rate": 0.0005, + "loss": 1.094, + "step": 525 + }, + { + "epoch": 0.57, + "learning_rate": 0.0005, + "loss": 1.1168, + "step": 526 + }, + { + "epoch": 0.57, + "learning_rate": 0.0005, + "loss": 1.1368, + "step": 527 + }, + { + "epoch": 0.57, + "learning_rate": 0.0005, + "loss": 1.1093, + "step": 528 + }, + { + "epoch": 0.58, + "learning_rate": 0.0005, + "loss": 1.1417, + "step": 529 + }, + { + "epoch": 0.58, + "learning_rate": 0.0005, + "loss": 1.1014, + "step": 530 + }, + { + "epoch": 0.58, + "learning_rate": 0.0005, + "loss": 1.1483, + "step": 531 + }, + { + "epoch": 0.58, + "learning_rate": 0.0005, + "loss": 1.1375, + "step": 532 + }, + { + "epoch": 0.58, + "learning_rate": 0.0005, + "loss": 1.1301, + "step": 533 + }, + { + "epoch": 0.58, + "learning_rate": 0.0005, + "loss": 1.1577, + "step": 534 + }, + { + "epoch": 0.58, + "learning_rate": 0.0005, + "loss": 1.1538, + "step": 535 + }, + { + "epoch": 0.58, + "learning_rate": 0.0005, + "loss": 1.1019, + "step": 536 + }, + { + "epoch": 0.58, + "learning_rate": 0.0005, + "loss": 1.0672, + "step": 537 + }, + { + "epoch": 0.59, + "learning_rate": 0.0005, + "loss": 1.1381, + "step": 538 + }, + { + "epoch": 0.59, + "learning_rate": 0.0005, + "loss": 1.1205, + "step": 539 + }, + { + "epoch": 0.59, + "learning_rate": 0.0005, + "loss": 1.1437, + "step": 540 + }, + { + "epoch": 0.59, + "learning_rate": 0.0005, + "loss": 1.1452, + "step": 541 + }, + { + "epoch": 0.59, + "learning_rate": 0.0005, + "loss": 1.1487, + "step": 542 + }, + { + "epoch": 0.59, + "learning_rate": 0.0005, + "loss": 1.1483, + "step": 543 + }, + { + "epoch": 0.59, + "learning_rate": 0.0005, + "loss": 1.1613, + "step": 544 + }, + { + "epoch": 0.59, + "learning_rate": 0.0005, + "loss": 1.171, + "step": 545 + }, + { + "epoch": 0.59, + "learning_rate": 0.0005, + "loss": 1.1422, + "step": 546 + }, + { + "epoch": 0.6, + "learning_rate": 0.0005, + "loss": 1.0819, + "step": 547 + }, + { + "epoch": 0.6, + "learning_rate": 0.0005, + "loss": 1.151, + "step": 548 + }, + { + "epoch": 0.6, + "learning_rate": 0.0005, + "loss": 1.1284, + "step": 549 + }, + { + "epoch": 0.6, + "learning_rate": 0.0005, + "loss": 1.0728, + "step": 550 + }, + { + "epoch": 0.6, + "learning_rate": 0.0005, + "loss": 1.1174, + "step": 551 + }, + { + "epoch": 0.6, + "learning_rate": 0.0005, + "loss": 1.0932, + "step": 552 + }, + { + "epoch": 0.6, + "learning_rate": 0.0005, + "loss": 1.1282, + "step": 553 + }, + { + "epoch": 0.6, + "learning_rate": 0.0005, + "loss": 1.1164, + "step": 554 + }, + { + "epoch": 0.6, + "learning_rate": 0.0005, + "loss": 1.1183, + "step": 555 + }, + { + "epoch": 0.61, + "learning_rate": 0.0005, + "loss": 1.1239, + "step": 556 + }, + { + "epoch": 0.61, + "learning_rate": 0.0005, + "loss": 1.158, + "step": 557 + }, + { + "epoch": 0.61, + "learning_rate": 0.0005, + "loss": 1.0952, + "step": 558 + }, + { + "epoch": 0.61, + "learning_rate": 0.0005, + "loss": 1.1344, + "step": 559 + }, + { + "epoch": 0.61, + "learning_rate": 0.0005, + "loss": 1.0638, + "step": 560 + }, + { + "epoch": 0.61, + "learning_rate": 0.0005, + "loss": 1.1002, + "step": 561 + }, + { + "epoch": 0.61, + "learning_rate": 0.0005, + "loss": 1.1048, + "step": 562 + }, + { + "epoch": 0.61, + "learning_rate": 0.0005, + "loss": 1.1261, + "step": 563 + }, + { + "epoch": 0.61, + "learning_rate": 0.0005, + "loss": 1.1568, + "step": 564 + }, + { + "epoch": 0.61, + "learning_rate": 0.0005, + "loss": 1.1359, + "step": 565 + }, + { + "epoch": 0.62, + "learning_rate": 0.0005, + "loss": 1.1488, + "step": 566 + }, + { + "epoch": 0.62, + "learning_rate": 0.0005, + "loss": 1.059, + "step": 567 + }, + { + "epoch": 0.62, + "learning_rate": 0.0005, + "loss": 1.1194, + "step": 568 + }, + { + "epoch": 0.62, + "learning_rate": 0.0005, + "loss": 1.0967, + "step": 569 + }, + { + "epoch": 0.62, + "learning_rate": 0.0005, + "loss": 1.1654, + "step": 570 + }, + { + "epoch": 0.62, + "learning_rate": 0.0005, + "loss": 1.1046, + "step": 571 + }, + { + "epoch": 0.62, + "learning_rate": 0.0005, + "loss": 1.1401, + "step": 572 + }, + { + "epoch": 0.62, + "learning_rate": 0.0005, + "loss": 1.1222, + "step": 573 + }, + { + "epoch": 0.62, + "learning_rate": 0.0005, + "loss": 1.1122, + "step": 574 + }, + { + "epoch": 0.63, + "learning_rate": 0.0005, + "loss": 1.088, + "step": 575 + }, + { + "epoch": 0.63, + "learning_rate": 0.0005, + "loss": 1.1294, + "step": 576 + }, + { + "epoch": 0.63, + "learning_rate": 0.0005, + "loss": 1.0765, + "step": 577 + }, + { + "epoch": 0.63, + "learning_rate": 0.0005, + "loss": 1.1138, + "step": 578 + }, + { + "epoch": 0.63, + "learning_rate": 0.0005, + "loss": 1.1306, + "step": 579 + }, + { + "epoch": 0.63, + "learning_rate": 0.0005, + "loss": 1.1253, + "step": 580 + }, + { + "epoch": 0.63, + "learning_rate": 0.0005, + "loss": 1.1401, + "step": 581 + }, + { + "epoch": 0.63, + "learning_rate": 0.0005, + "loss": 1.1404, + "step": 582 + }, + { + "epoch": 0.63, + "learning_rate": 0.0005, + "loss": 1.1131, + "step": 583 + }, + { + "epoch": 0.64, + "learning_rate": 0.0005, + "loss": 1.0914, + "step": 584 + }, + { + "epoch": 0.64, + "learning_rate": 0.0005, + "loss": 1.0828, + "step": 585 + }, + { + "epoch": 0.64, + "learning_rate": 0.0005, + "loss": 1.1405, + "step": 586 + }, + { + "epoch": 0.64, + "learning_rate": 0.0005, + "loss": 1.1837, + "step": 587 + }, + { + "epoch": 0.64, + "learning_rate": 0.0005, + "loss": 1.1506, + "step": 588 + }, + { + "epoch": 0.64, + "learning_rate": 0.0005, + "loss": 1.1738, + "step": 589 + }, + { + "epoch": 0.64, + "learning_rate": 0.0005, + "loss": 1.1206, + "step": 590 + }, + { + "epoch": 0.64, + "learning_rate": 0.0005, + "loss": 1.0608, + "step": 591 + }, + { + "epoch": 0.64, + "learning_rate": 0.0005, + "loss": 1.0809, + "step": 592 + }, + { + "epoch": 0.65, + "learning_rate": 0.0005, + "loss": 1.1042, + "step": 593 + }, + { + "epoch": 0.65, + "learning_rate": 0.0005, + "loss": 1.1272, + "step": 594 + }, + { + "epoch": 0.65, + "learning_rate": 0.0005, + "loss": 1.116, + "step": 595 + }, + { + "epoch": 0.65, + "learning_rate": 0.0005, + "loss": 1.1478, + "step": 596 + }, + { + "epoch": 0.65, + "learning_rate": 0.0005, + "loss": 1.1259, + "step": 597 + }, + { + "epoch": 0.65, + "learning_rate": 0.0005, + "loss": 1.1095, + "step": 598 + }, + { + "epoch": 0.65, + "learning_rate": 0.0005, + "loss": 1.1013, + "step": 599 + }, + { + "epoch": 0.65, + "learning_rate": 0.0005, + "loss": 1.0996, + "step": 600 + }, + { + "epoch": 0.65, + "learning_rate": 0.0005, + "loss": 1.1599, + "step": 601 + }, + { + "epoch": 0.66, + "learning_rate": 0.0005, + "loss": 1.0754, + "step": 602 + }, + { + "epoch": 0.66, + "learning_rate": 0.0005, + "loss": 1.1131, + "step": 603 + }, + { + "epoch": 0.66, + "learning_rate": 0.0005, + "loss": 1.1147, + "step": 604 + }, + { + "epoch": 0.66, + "learning_rate": 0.0005, + "loss": 1.0914, + "step": 605 + }, + { + "epoch": 0.66, + "learning_rate": 0.0005, + "loss": 1.1103, + "step": 606 + }, + { + "epoch": 0.66, + "learning_rate": 0.0005, + "loss": 1.1215, + "step": 607 + }, + { + "epoch": 0.66, + "learning_rate": 0.0005, + "loss": 1.1286, + "step": 608 + }, + { + "epoch": 0.66, + "learning_rate": 0.0005, + "loss": 1.1102, + "step": 609 + }, + { + "epoch": 0.66, + "learning_rate": 0.0005, + "loss": 1.1272, + "step": 610 + }, + { + "epoch": 0.66, + "learning_rate": 0.0005, + "loss": 1.1174, + "step": 611 + }, + { + "epoch": 0.67, + "learning_rate": 0.0005, + "loss": 1.1152, + "step": 612 + }, + { + "epoch": 0.67, + "learning_rate": 0.0005, + "loss": 1.138, + "step": 613 + }, + { + "epoch": 0.67, + "learning_rate": 0.0005, + "loss": 1.1373, + "step": 614 + }, + { + "epoch": 0.67, + "learning_rate": 0.0005, + "loss": 1.1462, + "step": 615 + }, + { + "epoch": 0.67, + "learning_rate": 0.0005, + "loss": 1.1455, + "step": 616 + }, + { + "epoch": 0.67, + "learning_rate": 0.0005, + "loss": 1.1221, + "step": 617 + }, + { + "epoch": 0.67, + "learning_rate": 0.0005, + "loss": 1.1226, + "step": 618 + }, + { + "epoch": 0.67, + "learning_rate": 0.0005, + "loss": 1.1146, + "step": 619 + }, + { + "epoch": 0.67, + "learning_rate": 0.0005, + "loss": 1.1333, + "step": 620 + }, + { + "epoch": 0.68, + "learning_rate": 0.0005, + "loss": 1.1346, + "step": 621 + }, + { + "epoch": 0.68, + "learning_rate": 0.0005, + "loss": 1.1345, + "step": 622 + }, + { + "epoch": 0.68, + "learning_rate": 0.0005, + "loss": 1.079, + "step": 623 + }, + { + "epoch": 0.68, + "learning_rate": 0.0005, + "loss": 1.0879, + "step": 624 + }, + { + "epoch": 0.68, + "learning_rate": 0.0005, + "loss": 1.1246, + "step": 625 + }, + { + "epoch": 0.68, + "learning_rate": 0.0005, + "loss": 1.1081, + "step": 626 + }, + { + "epoch": 0.68, + "learning_rate": 0.0005, + "loss": 1.1525, + "step": 627 + }, + { + "epoch": 0.68, + "learning_rate": 0.0005, + "loss": 1.0974, + "step": 628 + }, + { + "epoch": 0.68, + "learning_rate": 0.0005, + "loss": 1.1058, + "step": 629 + }, + { + "epoch": 0.69, + "learning_rate": 0.0005, + "loss": 1.065, + "step": 630 + }, + { + "epoch": 0.69, + "learning_rate": 0.0005, + "loss": 1.1341, + "step": 631 + }, + { + "epoch": 0.69, + "learning_rate": 0.0005, + "loss": 1.1271, + "step": 632 + }, + { + "epoch": 0.69, + "learning_rate": 0.0005, + "loss": 1.1319, + "step": 633 + }, + { + "epoch": 0.69, + "learning_rate": 0.0005, + "loss": 1.0693, + "step": 634 + }, + { + "epoch": 0.69, + "learning_rate": 0.0005, + "loss": 1.0998, + "step": 635 + }, + { + "epoch": 0.69, + "learning_rate": 0.0005, + "loss": 1.1202, + "step": 636 + }, + { + "epoch": 0.69, + "learning_rate": 0.0005, + "loss": 1.1288, + "step": 637 + }, + { + "epoch": 0.69, + "learning_rate": 0.0005, + "loss": 1.0911, + "step": 638 + }, + { + "epoch": 0.7, + "learning_rate": 0.0005, + "loss": 1.1047, + "step": 639 + }, + { + "epoch": 0.7, + "learning_rate": 0.0005, + "loss": 1.1118, + "step": 640 + }, + { + "epoch": 0.7, + "learning_rate": 0.0005, + "loss": 1.1363, + "step": 641 + }, + { + "epoch": 0.7, + "learning_rate": 0.0005, + "loss": 1.1063, + "step": 642 + }, + { + "epoch": 0.7, + "learning_rate": 0.0005, + "loss": 1.0746, + "step": 643 + }, + { + "epoch": 0.7, + "learning_rate": 0.0005, + "loss": 1.1267, + "step": 644 + }, + { + "epoch": 0.7, + "learning_rate": 0.0005, + "loss": 1.1338, + "step": 645 + }, + { + "epoch": 0.7, + "learning_rate": 0.0005, + "loss": 1.1241, + "step": 646 + }, + { + "epoch": 0.7, + "learning_rate": 0.0005, + "loss": 1.0913, + "step": 647 + }, + { + "epoch": 0.71, + "learning_rate": 0.0005, + "loss": 1.1116, + "step": 648 + }, + { + "epoch": 0.71, + "learning_rate": 0.0005, + "loss": 1.1093, + "step": 649 + }, + { + "epoch": 0.71, + "learning_rate": 0.0005, + "loss": 1.0802, + "step": 650 + }, + { + "epoch": 0.71, + "learning_rate": 0.0005, + "loss": 1.0975, + "step": 651 + }, + { + "epoch": 0.71, + "learning_rate": 0.0005, + "loss": 1.1244, + "step": 652 + }, + { + "epoch": 0.71, + "learning_rate": 0.0005, + "loss": 1.107, + "step": 653 + }, + { + "epoch": 0.71, + "learning_rate": 0.0005, + "loss": 1.074, + "step": 654 + }, + { + "epoch": 0.71, + "learning_rate": 0.0005, + "loss": 1.1133, + "step": 655 + }, + { + "epoch": 0.71, + "learning_rate": 0.0005, + "loss": 1.1446, + "step": 656 + }, + { + "epoch": 0.71, + "learning_rate": 0.0005, + "loss": 1.1164, + "step": 657 + }, + { + "epoch": 0.72, + "learning_rate": 0.0005, + "loss": 1.0897, + "step": 658 + }, + { + "epoch": 0.72, + "learning_rate": 0.0005, + "loss": 1.1515, + "step": 659 + }, + { + "epoch": 0.72, + "learning_rate": 0.0005, + "loss": 1.0802, + "step": 660 + }, + { + "epoch": 0.72, + "learning_rate": 0.0005, + "loss": 1.126, + "step": 661 + }, + { + "epoch": 0.72, + "learning_rate": 0.0005, + "loss": 1.1353, + "step": 662 + }, + { + "epoch": 0.72, + "learning_rate": 0.0005, + "loss": 1.0922, + "step": 663 + }, + { + "epoch": 0.72, + "learning_rate": 0.0005, + "loss": 1.0776, + "step": 664 + }, + { + "epoch": 0.72, + "learning_rate": 0.0005, + "loss": 1.1221, + "step": 665 + }, + { + "epoch": 0.72, + "learning_rate": 0.0005, + "loss": 1.0857, + "step": 666 + }, + { + "epoch": 0.73, + "learning_rate": 0.0005, + "loss": 1.0857, + "step": 667 + }, + { + "epoch": 0.73, + "learning_rate": 0.0005, + "loss": 1.068, + "step": 668 + }, + { + "epoch": 0.73, + "learning_rate": 0.0005, + "loss": 1.1664, + "step": 669 + }, + { + "epoch": 0.73, + "learning_rate": 0.0005, + "loss": 1.1275, + "step": 670 + }, + { + "epoch": 0.73, + "learning_rate": 0.0005, + "loss": 1.104, + "step": 671 + }, + { + "epoch": 0.73, + "learning_rate": 0.0005, + "loss": 1.0968, + "step": 672 + }, + { + "epoch": 0.73, + "learning_rate": 0.0005, + "loss": 1.0742, + "step": 673 + }, + { + "epoch": 0.73, + "learning_rate": 0.0005, + "loss": 1.1053, + "step": 674 + }, + { + "epoch": 0.73, + "learning_rate": 0.0005, + "loss": 1.098, + "step": 675 + }, + { + "epoch": 0.74, + "learning_rate": 0.0005, + "loss": 1.1067, + "step": 676 + }, + { + "epoch": 0.74, + "learning_rate": 0.0005, + "loss": 1.0613, + "step": 677 + }, + { + "epoch": 0.74, + "learning_rate": 0.0005, + "loss": 1.1308, + "step": 678 + }, + { + "epoch": 0.74, + "learning_rate": 0.0005, + "loss": 1.1962, + "step": 679 + }, + { + "epoch": 0.74, + "learning_rate": 0.0005, + "loss": 1.1398, + "step": 680 + }, + { + "epoch": 0.74, + "learning_rate": 0.0005, + "loss": 1.1324, + "step": 681 + }, + { + "epoch": 0.74, + "learning_rate": 0.0005, + "loss": 1.0799, + "step": 682 + }, + { + "epoch": 0.74, + "learning_rate": 0.0005, + "loss": 1.0756, + "step": 683 + }, + { + "epoch": 0.74, + "learning_rate": 0.0005, + "loss": 1.0795, + "step": 684 + }, + { + "epoch": 0.75, + "learning_rate": 0.0005, + "loss": 1.0653, + "step": 685 + }, + { + "epoch": 0.75, + "learning_rate": 0.0005, + "loss": 1.1439, + "step": 686 + }, + { + "epoch": 0.75, + "learning_rate": 0.0005, + "loss": 1.0828, + "step": 687 + }, + { + "epoch": 0.75, + "learning_rate": 0.0005, + "loss": 1.0688, + "step": 688 + }, + { + "epoch": 0.75, + "learning_rate": 0.0005, + "loss": 1.1265, + "step": 689 + }, + { + "epoch": 0.75, + "learning_rate": 0.0005, + "loss": 1.0815, + "step": 690 + }, + { + "epoch": 0.75, + "learning_rate": 0.0005, + "loss": 1.1125, + "step": 691 + }, + { + "epoch": 0.75, + "learning_rate": 0.0005, + "loss": 1.1091, + "step": 692 + }, + { + "epoch": 0.75, + "learning_rate": 0.0005, + "loss": 1.1174, + "step": 693 + }, + { + "epoch": 0.76, + "learning_rate": 0.0005, + "loss": 1.099, + "step": 694 + }, + { + "epoch": 0.76, + "learning_rate": 0.0005, + "loss": 1.1206, + "step": 695 + }, + { + "epoch": 0.76, + "learning_rate": 0.0005, + "loss": 1.0874, + "step": 696 + }, + { + "epoch": 0.76, + "learning_rate": 0.0005, + "loss": 1.1044, + "step": 697 + }, + { + "epoch": 0.76, + "learning_rate": 0.0005, + "loss": 1.0745, + "step": 698 + }, + { + "epoch": 0.76, + "learning_rate": 0.0005, + "loss": 1.0868, + "step": 699 + }, + { + "epoch": 0.76, + "learning_rate": 0.0005, + "loss": 1.1438, + "step": 700 + }, + { + "epoch": 0.76, + "learning_rate": 0.0005, + "loss": 1.1195, + "step": 701 + }, + { + "epoch": 0.76, + "learning_rate": 0.0005, + "loss": 1.1014, + "step": 702 + }, + { + "epoch": 0.76, + "learning_rate": 0.0005, + "loss": 1.1201, + "step": 703 + }, + { + "epoch": 0.77, + "learning_rate": 0.0005, + "loss": 1.0637, + "step": 704 + }, + { + "epoch": 0.77, + "learning_rate": 0.0005, + "loss": 1.0999, + "step": 705 + }, + { + "epoch": 0.77, + "learning_rate": 0.0005, + "loss": 1.0813, + "step": 706 + }, + { + "epoch": 0.77, + "learning_rate": 0.0005, + "loss": 1.1071, + "step": 707 + }, + { + "epoch": 0.77, + "learning_rate": 0.0005, + "loss": 1.1175, + "step": 708 + }, + { + "epoch": 0.77, + "learning_rate": 0.0005, + "loss": 1.0857, + "step": 709 + }, + { + "epoch": 0.77, + "learning_rate": 0.0005, + "loss": 1.1159, + "step": 710 + }, + { + "epoch": 0.77, + "learning_rate": 0.0005, + "loss": 1.1585, + "step": 711 + }, + { + "epoch": 0.77, + "learning_rate": 0.0005, + "loss": 1.1417, + "step": 712 + }, + { + "epoch": 0.78, + "learning_rate": 0.0005, + "loss": 1.0656, + "step": 713 + }, + { + "epoch": 0.78, + "learning_rate": 0.0005, + "loss": 1.1289, + "step": 714 + }, + { + "epoch": 0.78, + "learning_rate": 0.0005, + "loss": 1.1022, + "step": 715 + }, + { + "epoch": 0.78, + "learning_rate": 0.0005, + "loss": 1.1085, + "step": 716 + }, + { + "epoch": 0.78, + "learning_rate": 0.0005, + "loss": 1.0929, + "step": 717 + }, + { + "epoch": 0.78, + "learning_rate": 0.0005, + "loss": 1.1263, + "step": 718 + }, + { + "epoch": 0.78, + "learning_rate": 0.0005, + "loss": 1.1048, + "step": 719 + }, + { + "epoch": 0.78, + "learning_rate": 0.0005, + "loss": 1.1229, + "step": 720 + }, + { + "epoch": 0.78, + "learning_rate": 0.0005, + "loss": 1.0817, + "step": 721 + }, + { + "epoch": 0.79, + "learning_rate": 0.0005, + "loss": 1.1005, + "step": 722 + }, + { + "epoch": 0.79, + "learning_rate": 0.0005, + "loss": 1.0963, + "step": 723 + }, + { + "epoch": 0.79, + "learning_rate": 0.0005, + "loss": 1.1059, + "step": 724 + }, + { + "epoch": 0.79, + "learning_rate": 0.0005, + "loss": 1.0775, + "step": 725 + }, + { + "epoch": 0.79, + "learning_rate": 0.0005, + "loss": 1.0943, + "step": 726 + }, + { + "epoch": 0.79, + "learning_rate": 0.0005, + "loss": 1.132, + "step": 727 + }, + { + "epoch": 0.79, + "learning_rate": 0.0005, + "loss": 1.1033, + "step": 728 + }, + { + "epoch": 0.79, + "learning_rate": 0.0005, + "loss": 1.115, + "step": 729 + }, + { + "epoch": 0.79, + "learning_rate": 0.0005, + "loss": 1.1164, + "step": 730 + }, + { + "epoch": 0.8, + "learning_rate": 0.0005, + "loss": 1.0839, + "step": 731 + }, + { + "epoch": 0.8, + "learning_rate": 0.0005, + "loss": 1.079, + "step": 732 + }, + { + "epoch": 0.8, + "learning_rate": 0.0005, + "loss": 1.1161, + "step": 733 + }, + { + "epoch": 0.8, + "learning_rate": 0.0005, + "loss": 1.0919, + "step": 734 + }, + { + "epoch": 0.8, + "learning_rate": 0.0005, + "loss": 1.1426, + "step": 735 + }, + { + "epoch": 0.8, + "learning_rate": 0.0005, + "loss": 1.1042, + "step": 736 + }, + { + "epoch": 0.8, + "learning_rate": 0.0005, + "loss": 1.1555, + "step": 737 + }, + { + "epoch": 0.8, + "learning_rate": 0.0005, + "loss": 1.0744, + "step": 738 + }, + { + "epoch": 0.8, + "learning_rate": 0.0005, + "loss": 1.0961, + "step": 739 + }, + { + "epoch": 0.81, + "learning_rate": 0.0005, + "loss": 1.087, + "step": 740 + }, + { + "epoch": 0.81, + "learning_rate": 0.0005, + "loss": 1.1187, + "step": 741 + }, + { + "epoch": 0.81, + "learning_rate": 0.0005, + "loss": 1.0824, + "step": 742 + }, + { + "epoch": 0.81, + "learning_rate": 0.0005, + "loss": 1.1511, + "step": 743 + }, + { + "epoch": 0.81, + "learning_rate": 0.0005, + "loss": 1.0868, + "step": 744 + }, + { + "epoch": 0.81, + "learning_rate": 0.0005, + "loss": 1.1074, + "step": 745 + }, + { + "epoch": 0.81, + "learning_rate": 0.0005, + "loss": 1.0735, + "step": 746 + }, + { + "epoch": 0.81, + "learning_rate": 0.0005, + "loss": 1.0966, + "step": 747 + }, + { + "epoch": 0.81, + "learning_rate": 0.0005, + "loss": 1.1027, + "step": 748 + }, + { + "epoch": 0.82, + "learning_rate": 0.0005, + "loss": 1.1406, + "step": 749 + }, + { + "epoch": 0.82, + "learning_rate": 0.0005, + "loss": 1.1066, + "step": 750 + }, + { + "epoch": 0.82, + "learning_rate": 0.0005, + "loss": 1.1331, + "step": 751 + }, + { + "epoch": 0.82, + "learning_rate": 0.0005, + "loss": 1.1133, + "step": 752 + }, + { + "epoch": 0.82, + "learning_rate": 0.0005, + "loss": 1.0843, + "step": 753 + }, + { + "epoch": 0.82, + "learning_rate": 0.0005, + "loss": 1.1669, + "step": 754 + }, + { + "epoch": 0.82, + "learning_rate": 0.0005, + "loss": 1.1684, + "step": 755 + }, + { + "epoch": 0.82, + "learning_rate": 0.0005, + "loss": 1.1325, + "step": 756 + }, + { + "epoch": 0.82, + "learning_rate": 0.0005, + "loss": 1.1278, + "step": 757 + }, + { + "epoch": 0.82, + "learning_rate": 0.0005, + "loss": 1.1433, + "step": 758 + }, + { + "epoch": 0.83, + "learning_rate": 0.0005, + "loss": 1.1071, + "step": 759 + }, + { + "epoch": 0.83, + "learning_rate": 0.0005, + "loss": 1.0884, + "step": 760 + }, + { + "epoch": 0.83, + "learning_rate": 0.0005, + "loss": 1.1432, + "step": 761 + }, + { + "epoch": 0.83, + "learning_rate": 0.0005, + "loss": 1.1135, + "step": 762 + }, + { + "epoch": 0.83, + "learning_rate": 0.0005, + "loss": 1.0831, + "step": 763 + }, + { + "epoch": 0.83, + "learning_rate": 0.0005, + "loss": 1.0794, + "step": 764 + }, + { + "epoch": 0.83, + "learning_rate": 0.0005, + "loss": 1.1232, + "step": 765 + }, + { + "epoch": 0.83, + "learning_rate": 0.0005, + "loss": 1.0984, + "step": 766 + }, + { + "epoch": 0.83, + "learning_rate": 0.0005, + "loss": 1.0881, + "step": 767 + }, + { + "epoch": 0.84, + "learning_rate": 0.0005, + "loss": 1.0875, + "step": 768 + }, + { + "epoch": 0.84, + "learning_rate": 0.0005, + "loss": 1.1447, + "step": 769 + }, + { + "epoch": 0.84, + "learning_rate": 0.0005, + "loss": 1.1168, + "step": 770 + }, + { + "epoch": 0.84, + "learning_rate": 0.0005, + "loss": 1.1088, + "step": 771 + }, + { + "epoch": 0.84, + "learning_rate": 0.0005, + "loss": 1.0764, + "step": 772 + }, + { + "epoch": 0.84, + "learning_rate": 0.0005, + "loss": 1.0941, + "step": 773 + }, + { + "epoch": 0.84, + "learning_rate": 0.0005, + "loss": 1.1003, + "step": 774 + }, + { + "epoch": 0.84, + "learning_rate": 0.0005, + "loss": 1.0956, + "step": 775 + }, + { + "epoch": 0.84, + "learning_rate": 0.0005, + "loss": 1.0576, + "step": 776 + }, + { + "epoch": 0.85, + "learning_rate": 0.0005, + "loss": 1.1346, + "step": 777 + }, + { + "epoch": 0.85, + "learning_rate": 0.0005, + "loss": 1.1333, + "step": 778 + }, + { + "epoch": 0.85, + "learning_rate": 0.0005, + "loss": 1.1416, + "step": 779 + }, + { + "epoch": 0.85, + "learning_rate": 0.0005, + "loss": 1.1313, + "step": 780 + }, + { + "epoch": 0.85, + "learning_rate": 0.0005, + "loss": 1.0697, + "step": 781 + }, + { + "epoch": 0.85, + "learning_rate": 0.0005, + "loss": 1.1134, + "step": 782 + }, + { + "epoch": 0.85, + "learning_rate": 0.0005, + "loss": 1.1372, + "step": 783 + }, + { + "epoch": 0.85, + "learning_rate": 0.0005, + "loss": 1.1014, + "step": 784 + }, + { + "epoch": 0.85, + "learning_rate": 0.0005, + "loss": 1.0818, + "step": 785 + }, + { + "epoch": 0.86, + "learning_rate": 0.0005, + "loss": 1.123, + "step": 786 + }, + { + "epoch": 0.86, + "learning_rate": 0.0005, + "loss": 1.1136, + "step": 787 + }, + { + "epoch": 0.86, + "learning_rate": 0.0005, + "loss": 1.1025, + "step": 788 + }, + { + "epoch": 0.86, + "learning_rate": 0.0005, + "loss": 1.0854, + "step": 789 + }, + { + "epoch": 0.86, + "learning_rate": 0.0005, + "loss": 1.0855, + "step": 790 + }, + { + "epoch": 0.86, + "learning_rate": 0.0005, + "loss": 1.0953, + "step": 791 + }, + { + "epoch": 0.86, + "learning_rate": 0.0005, + "loss": 1.1055, + "step": 792 + }, + { + "epoch": 0.86, + "learning_rate": 0.0005, + "loss": 1.1145, + "step": 793 + }, + { + "epoch": 0.86, + "learning_rate": 0.0005, + "loss": 1.0638, + "step": 794 + }, + { + "epoch": 0.87, + "learning_rate": 0.0005, + "loss": 1.1167, + "step": 795 + }, + { + "epoch": 0.87, + "learning_rate": 0.0005, + "loss": 1.099, + "step": 796 + }, + { + "epoch": 0.87, + "learning_rate": 0.0005, + "loss": 1.1113, + "step": 797 + }, + { + "epoch": 0.87, + "learning_rate": 0.0005, + "loss": 1.0363, + "step": 798 + }, + { + "epoch": 0.87, + "learning_rate": 0.0005, + "loss": 1.1093, + "step": 799 + }, + { + "epoch": 0.87, + "learning_rate": 0.0005, + "loss": 1.0849, + "step": 800 + }, + { + "epoch": 0.87, + "learning_rate": 0.0005, + "loss": 1.1062, + "step": 801 + }, + { + "epoch": 0.87, + "learning_rate": 0.0005, + "loss": 1.0502, + "step": 802 + }, + { + "epoch": 0.87, + "learning_rate": 0.0005, + "loss": 1.1058, + "step": 803 + }, + { + "epoch": 0.87, + "learning_rate": 0.0005, + "loss": 1.1806, + "step": 804 + }, + { + "epoch": 0.88, + "learning_rate": 0.0005, + "loss": 1.1451, + "step": 805 + }, + { + "epoch": 0.88, + "learning_rate": 0.0005, + "loss": 1.0792, + "step": 806 + }, + { + "epoch": 0.88, + "learning_rate": 0.0005, + "loss": 1.0959, + "step": 807 + }, + { + "epoch": 0.88, + "learning_rate": 0.0005, + "loss": 1.1004, + "step": 808 + }, + { + "epoch": 0.88, + "learning_rate": 0.0005, + "loss": 1.1143, + "step": 809 + }, + { + "epoch": 0.88, + "learning_rate": 0.0005, + "loss": 1.1294, + "step": 810 + }, + { + "epoch": 0.88, + "learning_rate": 0.0005, + "loss": 1.1008, + "step": 811 + }, + { + "epoch": 0.88, + "learning_rate": 0.0005, + "loss": 1.1182, + "step": 812 + }, + { + "epoch": 0.88, + "learning_rate": 0.0005, + "loss": 1.117, + "step": 813 + }, + { + "epoch": 0.89, + "learning_rate": 0.0005, + "loss": 1.1207, + "step": 814 + }, + { + "epoch": 0.89, + "learning_rate": 0.0005, + "loss": 1.1213, + "step": 815 + }, + { + "epoch": 0.89, + "learning_rate": 0.0005, + "loss": 1.0676, + "step": 816 + }, + { + "epoch": 0.89, + "learning_rate": 0.0005, + "loss": 1.1163, + "step": 817 + }, + { + "epoch": 0.89, + "learning_rate": 0.0005, + "loss": 1.0947, + "step": 818 + }, + { + "epoch": 0.89, + "learning_rate": 0.0005, + "loss": 1.1144, + "step": 819 + }, + { + "epoch": 0.89, + "learning_rate": 0.0005, + "loss": 1.0878, + "step": 820 + }, + { + "epoch": 0.89, + "learning_rate": 0.0005, + "loss": 1.1174, + "step": 821 + }, + { + "epoch": 0.89, + "learning_rate": 0.0005, + "loss": 1.0926, + "step": 822 + }, + { + "epoch": 0.9, + "learning_rate": 0.0005, + "loss": 1.1066, + "step": 823 + }, + { + "epoch": 0.9, + "learning_rate": 0.0005, + "loss": 1.1115, + "step": 824 + }, + { + "epoch": 0.9, + "learning_rate": 0.0005, + "loss": 1.0585, + "step": 825 + }, + { + "epoch": 0.9, + "learning_rate": 0.0005, + "loss": 1.1201, + "step": 826 + }, + { + "epoch": 0.9, + "learning_rate": 0.0005, + "loss": 1.0767, + "step": 827 + }, + { + "epoch": 0.9, + "learning_rate": 0.0005, + "loss": 1.0875, + "step": 828 + }, + { + "epoch": 0.9, + "learning_rate": 0.0005, + "loss": 1.0978, + "step": 829 + }, + { + "epoch": 0.9, + "learning_rate": 0.0005, + "loss": 1.1338, + "step": 830 + }, + { + "epoch": 0.9, + "learning_rate": 0.0005, + "loss": 1.108, + "step": 831 + }, + { + "epoch": 0.91, + "learning_rate": 0.0005, + "loss": 1.0795, + "step": 832 + }, + { + "epoch": 0.91, + "learning_rate": 0.0005, + "loss": 1.11, + "step": 833 + }, + { + "epoch": 0.91, + "learning_rate": 0.0005, + "loss": 1.0496, + "step": 834 + }, + { + "epoch": 0.91, + "learning_rate": 0.0005, + "loss": 1.0698, + "step": 835 + }, + { + "epoch": 0.91, + "learning_rate": 0.0005, + "loss": 1.099, + "step": 836 + }, + { + "epoch": 0.91, + "learning_rate": 0.0005, + "loss": 1.1051, + "step": 837 + }, + { + "epoch": 0.91, + "learning_rate": 0.0005, + "loss": 1.0613, + "step": 838 + }, + { + "epoch": 0.91, + "learning_rate": 0.0005, + "loss": 1.116, + "step": 839 + }, + { + "epoch": 0.91, + "learning_rate": 0.0005, + "loss": 1.0896, + "step": 840 + }, + { + "epoch": 0.92, + "learning_rate": 0.0005, + "loss": 1.1035, + "step": 841 + }, + { + "epoch": 0.92, + "learning_rate": 0.0005, + "loss": 1.0452, + "step": 842 + }, + { + "epoch": 0.92, + "learning_rate": 0.0005, + "loss": 1.0899, + "step": 843 + }, + { + "epoch": 0.92, + "learning_rate": 0.0005, + "loss": 1.1036, + "step": 844 + }, + { + "epoch": 0.92, + "learning_rate": 0.0005, + "loss": 1.1196, + "step": 845 + }, + { + "epoch": 0.92, + "learning_rate": 0.0005, + "loss": 1.0674, + "step": 846 + }, + { + "epoch": 0.92, + "learning_rate": 0.0005, + "loss": 1.0857, + "step": 847 + }, + { + "epoch": 0.92, + "learning_rate": 0.0005, + "loss": 1.0712, + "step": 848 + }, + { + "epoch": 0.92, + "learning_rate": 0.0005, + "loss": 1.1294, + "step": 849 + }, + { + "epoch": 0.92, + "learning_rate": 0.0005, + "loss": 1.0668, + "step": 850 + }, + { + "epoch": 0.93, + "learning_rate": 0.0005, + "loss": 1.0598, + "step": 851 + }, + { + "epoch": 0.93, + "learning_rate": 0.0005, + "loss": 1.1109, + "step": 852 + }, + { + "epoch": 0.93, + "learning_rate": 0.0005, + "loss": 1.0763, + "step": 853 + }, + { + "epoch": 0.93, + "learning_rate": 0.0005, + "loss": 1.1044, + "step": 854 + }, + { + "epoch": 0.93, + "learning_rate": 0.0005, + "loss": 1.1576, + "step": 855 + }, + { + "epoch": 0.93, + "learning_rate": 0.0005, + "loss": 1.1349, + "step": 856 + }, + { + "epoch": 0.93, + "learning_rate": 0.0005, + "loss": 1.1037, + "step": 857 + }, + { + "epoch": 0.93, + "learning_rate": 0.0005, + "loss": 1.1247, + "step": 858 + }, + { + "epoch": 0.93, + "learning_rate": 0.0005, + "loss": 1.1302, + "step": 859 + }, + { + "epoch": 0.94, + "learning_rate": 0.0005, + "loss": 1.1072, + "step": 860 + }, + { + "epoch": 0.94, + "learning_rate": 0.0005, + "loss": 1.0663, + "step": 861 + }, + { + "epoch": 0.94, + "learning_rate": 0.0005, + "loss": 1.119, + "step": 862 + }, + { + "epoch": 0.94, + "learning_rate": 0.0005, + "loss": 1.1027, + "step": 863 + }, + { + "epoch": 0.94, + "learning_rate": 0.0005, + "loss": 1.1633, + "step": 864 + }, + { + "epoch": 0.94, + "learning_rate": 0.0005, + "loss": 1.0857, + "step": 865 + }, + { + "epoch": 0.94, + "learning_rate": 0.0005, + "loss": 1.1031, + "step": 866 + }, + { + "epoch": 0.94, + "learning_rate": 0.0005, + "loss": 1.0759, + "step": 867 + }, + { + "epoch": 0.94, + "learning_rate": 0.0005, + "loss": 1.0975, + "step": 868 + }, + { + "epoch": 0.95, + "learning_rate": 0.0005, + "loss": 1.1353, + "step": 869 + }, + { + "epoch": 0.95, + "learning_rate": 0.0005, + "loss": 1.0713, + "step": 870 + }, + { + "epoch": 0.95, + "learning_rate": 0.0005, + "loss": 1.1092, + "step": 871 + }, + { + "epoch": 0.95, + "learning_rate": 0.0005, + "loss": 1.1258, + "step": 872 + }, + { + "epoch": 0.95, + "learning_rate": 0.0005, + "loss": 1.0857, + "step": 873 + }, + { + "epoch": 0.95, + "learning_rate": 0.0005, + "loss": 1.0982, + "step": 874 + }, + { + "epoch": 0.95, + "learning_rate": 0.0005, + "loss": 1.0857, + "step": 875 + }, + { + "epoch": 0.95, + "learning_rate": 0.0005, + "loss": 1.047, + "step": 876 + }, + { + "epoch": 0.95, + "learning_rate": 0.0005, + "loss": 1.0826, + "step": 877 + }, + { + "epoch": 0.96, + "learning_rate": 0.0005, + "loss": 1.0755, + "step": 878 + }, + { + "epoch": 0.96, + "learning_rate": 0.0005, + "loss": 1.1544, + "step": 879 + }, + { + "epoch": 0.96, + "learning_rate": 0.0005, + "loss": 1.1101, + "step": 880 + }, + { + "epoch": 0.96, + "learning_rate": 0.0005, + "loss": 1.0808, + "step": 881 + }, + { + "epoch": 0.96, + "learning_rate": 0.0005, + "loss": 1.0906, + "step": 882 + }, + { + "epoch": 0.96, + "learning_rate": 0.0005, + "loss": 1.0888, + "step": 883 + }, + { + "epoch": 0.96, + "learning_rate": 0.0005, + "loss": 1.0788, + "step": 884 + }, + { + "epoch": 0.96, + "learning_rate": 0.0005, + "loss": 1.0749, + "step": 885 + }, + { + "epoch": 0.96, + "learning_rate": 0.0005, + "loss": 1.0795, + "step": 886 + }, + { + "epoch": 0.97, + "learning_rate": 0.0005, + "loss": 1.0649, + "step": 887 + }, + { + "epoch": 0.97, + "learning_rate": 0.0005, + "loss": 1.067, + "step": 888 + }, + { + "epoch": 0.97, + "learning_rate": 0.0005, + "loss": 1.0922, + "step": 889 + }, + { + "epoch": 0.97, + "learning_rate": 0.0005, + "loss": 1.1163, + "step": 890 + }, + { + "epoch": 0.97, + "learning_rate": 0.0005, + "loss": 1.0979, + "step": 891 + }, + { + "epoch": 0.97, + "learning_rate": 0.0005, + "loss": 1.0719, + "step": 892 + }, + { + "epoch": 0.97, + "learning_rate": 0.0005, + "loss": 1.0796, + "step": 893 + }, + { + "epoch": 0.97, + "learning_rate": 0.0005, + "loss": 1.1005, + "step": 894 + }, + { + "epoch": 0.97, + "learning_rate": 0.0005, + "loss": 1.0766, + "step": 895 + }, + { + "epoch": 0.97, + "learning_rate": 0.0005, + "loss": 1.1292, + "step": 896 + }, + { + "epoch": 0.98, + "learning_rate": 0.0005, + "loss": 1.1281, + "step": 897 + }, + { + "epoch": 0.98, + "learning_rate": 0.0005, + "loss": 1.0687, + "step": 898 + }, + { + "epoch": 0.98, + "learning_rate": 0.0005, + "loss": 1.0858, + "step": 899 + }, + { + "epoch": 0.98, + "learning_rate": 0.0005, + "loss": 1.0695, + "step": 900 + }, + { + "epoch": 0.98, + "learning_rate": 0.0005, + "loss": 1.0969, + "step": 901 + }, + { + "epoch": 0.98, + "learning_rate": 0.0005, + "loss": 1.0681, + "step": 902 + }, + { + "epoch": 0.98, + "learning_rate": 0.0005, + "loss": 1.1084, + "step": 903 + }, + { + "epoch": 0.98, + "learning_rate": 0.0005, + "loss": 1.0828, + "step": 904 + }, + { + "epoch": 0.98, + "learning_rate": 0.0005, + "loss": 1.1158, + "step": 905 + }, + { + "epoch": 0.99, + "learning_rate": 0.0005, + "loss": 1.1029, + "step": 906 + }, + { + "epoch": 0.99, + "learning_rate": 0.0005, + "loss": 1.0668, + "step": 907 + }, + { + "epoch": 0.99, + "learning_rate": 0.0005, + "loss": 1.0865, + "step": 908 + }, + { + "epoch": 0.99, + "learning_rate": 0.0005, + "loss": 1.0623, + "step": 909 + }, + { + "epoch": 0.99, + "learning_rate": 0.0005, + "loss": 1.0685, + "step": 910 + }, + { + "epoch": 0.99, + "learning_rate": 0.0005, + "loss": 1.144, + "step": 911 + }, + { + "epoch": 0.99, + "learning_rate": 0.0005, + "loss": 1.1082, + "step": 912 + }, + { + "epoch": 0.99, + "learning_rate": 0.0005, + "loss": 1.1145, + "step": 913 + }, + { + "epoch": 0.99, + "learning_rate": 0.0005, + "loss": 1.1184, + "step": 914 + }, + { + "epoch": 1.0, + "learning_rate": 0.0005, + "loss": 1.0921, + "step": 915 + }, + { + "epoch": 1.0, + "learning_rate": 0.0005, + "loss": 1.0661, + "step": 916 + }, + { + "epoch": 1.0, + "learning_rate": 0.0005, + "loss": 1.0955, + "step": 917 + }, + { + "epoch": 1.0, + "learning_rate": 0.0005, + "loss": 1.114, + "step": 918 + }, + { + "epoch": 1.0, + "learning_rate": 0.0005, + "loss": 1.2113, + "step": 919 + }, + { + "epoch": 1.0, + "learning_rate": 0.0005, + "loss": 1.0899, + "step": 920 + }, + { + "epoch": 1.0, + "learning_rate": 0.0005, + "loss": 1.0396, + "step": 921 + }, + { + "epoch": 1.0, + "learning_rate": 0.0005, + "loss": 1.0667, + "step": 922 + }, + { + "epoch": 1.0, + "learning_rate": 0.0005, + "loss": 1.0786, + "step": 923 + }, + { + "epoch": 1.01, + "learning_rate": 0.0005, + "loss": 1.1367, + "step": 924 + }, + { + "epoch": 1.01, + "learning_rate": 0.0005, + "loss": 1.0459, + "step": 925 + }, + { + "epoch": 1.01, + "learning_rate": 0.0005, + "loss": 1.1318, + "step": 926 + }, + { + "epoch": 1.01, + "learning_rate": 0.0005, + "loss": 1.1005, + "step": 927 + }, + { + "epoch": 1.01, + "learning_rate": 0.0005, + "loss": 1.104, + "step": 928 + }, + { + "epoch": 1.01, + "learning_rate": 0.0005, + "loss": 1.035, + "step": 929 + }, + { + "epoch": 1.01, + "learning_rate": 0.0005, + "loss": 1.0806, + "step": 930 + }, + { + "epoch": 1.01, + "learning_rate": 0.0005, + "loss": 1.1043, + "step": 931 + }, + { + "epoch": 1.01, + "learning_rate": 0.0005, + "loss": 1.0933, + "step": 932 + }, + { + "epoch": 1.02, + "learning_rate": 0.0005, + "loss": 1.0306, + "step": 933 + }, + { + "epoch": 1.02, + "learning_rate": 0.0005, + "loss": 1.0372, + "step": 934 + }, + { + "epoch": 1.02, + "learning_rate": 0.0005, + "loss": 1.0866, + "step": 935 + }, + { + "epoch": 1.02, + "learning_rate": 0.0005, + "loss": 1.015, + "step": 936 + }, + { + "epoch": 1.02, + "learning_rate": 0.0005, + "loss": 1.0392, + "step": 937 + }, + { + "epoch": 1.02, + "learning_rate": 0.0005, + "loss": 1.0252, + "step": 938 + }, + { + "epoch": 1.02, + "learning_rate": 0.0005, + "loss": 1.0777, + "step": 939 + }, + { + "epoch": 1.02, + "learning_rate": 0.0005, + "loss": 1.0615, + "step": 940 + }, + { + "epoch": 1.02, + "learning_rate": 0.0005, + "loss": 1.088, + "step": 941 + }, + { + "epoch": 1.03, + "learning_rate": 0.0005, + "loss": 1.1266, + "step": 942 + }, + { + "epoch": 1.03, + "learning_rate": 0.0005, + "loss": 1.0666, + "step": 943 + }, + { + "epoch": 1.03, + "learning_rate": 0.0005, + "loss": 1.0538, + "step": 944 + }, + { + "epoch": 1.03, + "learning_rate": 0.0005, + "loss": 1.109, + "step": 945 + }, + { + "epoch": 1.03, + "learning_rate": 0.0005, + "loss": 1.0926, + "step": 946 + }, + { + "epoch": 1.03, + "learning_rate": 0.0005, + "loss": 1.0684, + "step": 947 + }, + { + "epoch": 1.03, + "learning_rate": 0.0005, + "loss": 1.0929, + "step": 948 + }, + { + "epoch": 1.03, + "learning_rate": 0.0005, + "loss": 1.0709, + "step": 949 + }, + { + "epoch": 1.03, + "learning_rate": 0.0005, + "loss": 1.0519, + "step": 950 + }, + { + "epoch": 1.03, + "learning_rate": 0.0005, + "loss": 1.0339, + "step": 951 + }, + { + "epoch": 1.04, + "learning_rate": 0.0005, + "loss": 1.0779, + "step": 952 + }, + { + "epoch": 1.04, + "learning_rate": 0.0005, + "loss": 1.0931, + "step": 953 + }, + { + "epoch": 1.04, + "learning_rate": 0.0005, + "loss": 1.1054, + "step": 954 + }, + { + "epoch": 1.04, + "learning_rate": 0.0005, + "loss": 1.0382, + "step": 955 + }, + { + "epoch": 1.04, + "learning_rate": 0.0005, + "loss": 1.0596, + "step": 956 + }, + { + "epoch": 1.04, + "learning_rate": 0.0005, + "loss": 1.101, + "step": 957 + }, + { + "epoch": 1.04, + "learning_rate": 0.0005, + "loss": 1.0891, + "step": 958 + }, + { + "epoch": 1.04, + "learning_rate": 0.0005, + "loss": 1.0377, + "step": 959 + }, + { + "epoch": 1.04, + "learning_rate": 0.0005, + "loss": 1.0933, + "step": 960 + }, + { + "epoch": 1.05, + "learning_rate": 0.0005, + "loss": 1.0286, + "step": 961 + }, + { + "epoch": 1.05, + "learning_rate": 0.0005, + "loss": 1.105, + "step": 962 + }, + { + "epoch": 1.05, + "learning_rate": 0.0005, + "loss": 1.0875, + "step": 963 + }, + { + "epoch": 1.05, + "learning_rate": 0.0005, + "loss": 1.1053, + "step": 964 + }, + { + "epoch": 1.05, + "learning_rate": 0.0005, + "loss": 1.0932, + "step": 965 + }, + { + "epoch": 1.05, + "learning_rate": 0.0005, + "loss": 1.028, + "step": 966 + }, + { + "epoch": 1.05, + "learning_rate": 0.0005, + "loss": 1.0649, + "step": 967 + }, + { + "epoch": 1.05, + "learning_rate": 0.0005, + "loss": 1.0729, + "step": 968 + }, + { + "epoch": 1.05, + "learning_rate": 0.0005, + "loss": 1.0475, + "step": 969 + }, + { + "epoch": 1.06, + "learning_rate": 0.0005, + "loss": 1.0725, + "step": 970 + }, + { + "epoch": 1.06, + "learning_rate": 0.0005, + "loss": 1.0785, + "step": 971 + }, + { + "epoch": 1.06, + "learning_rate": 0.0005, + "loss": 1.0486, + "step": 972 + }, + { + "epoch": 1.06, + "learning_rate": 0.0005, + "loss": 1.0837, + "step": 973 + }, + { + "epoch": 1.06, + "learning_rate": 0.0005, + "loss": 1.0987, + "step": 974 + }, + { + "epoch": 1.06, + "learning_rate": 0.0005, + "loss": 1.108, + "step": 975 + }, + { + "epoch": 1.06, + "learning_rate": 0.0005, + "loss": 1.0849, + "step": 976 + }, + { + "epoch": 1.06, + "learning_rate": 0.0005, + "loss": 1.0709, + "step": 977 + }, + { + "epoch": 1.06, + "learning_rate": 0.0005, + "loss": 1.092, + "step": 978 + }, + { + "epoch": 1.07, + "learning_rate": 0.0005, + "loss": 1.0899, + "step": 979 + }, + { + "epoch": 1.07, + "learning_rate": 0.0005, + "loss": 1.0819, + "step": 980 + }, + { + "epoch": 1.07, + "learning_rate": 0.0005, + "loss": 1.0788, + "step": 981 + }, + { + "epoch": 1.07, + "learning_rate": 0.0005, + "loss": 1.0674, + "step": 982 + }, + { + "epoch": 1.07, + "learning_rate": 0.0005, + "loss": 1.0349, + "step": 983 + }, + { + "epoch": 1.07, + "learning_rate": 0.0005, + "loss": 1.0599, + "step": 984 + }, + { + "epoch": 1.07, + "learning_rate": 0.0005, + "loss": 1.1402, + "step": 985 + }, + { + "epoch": 1.07, + "learning_rate": 0.0005, + "loss": 1.0974, + "step": 986 + }, + { + "epoch": 1.07, + "learning_rate": 0.0005, + "loss": 1.049, + "step": 987 + }, + { + "epoch": 1.08, + "learning_rate": 0.0005, + "loss": 1.0569, + "step": 988 + }, + { + "epoch": 1.08, + "learning_rate": 0.0005, + "loss": 1.0777, + "step": 989 + }, + { + "epoch": 1.08, + "learning_rate": 0.0005, + "loss": 1.1012, + "step": 990 + }, + { + "epoch": 1.08, + "learning_rate": 0.0005, + "loss": 1.0673, + "step": 991 + }, + { + "epoch": 1.08, + "learning_rate": 0.0005, + "loss": 1.0751, + "step": 992 + }, + { + "epoch": 1.08, + "learning_rate": 0.0005, + "loss": 1.1446, + "step": 993 + }, + { + "epoch": 1.08, + "learning_rate": 0.0005, + "loss": 1.082, + "step": 994 + }, + { + "epoch": 1.08, + "learning_rate": 0.0005, + "loss": 1.094, + "step": 995 + }, + { + "epoch": 1.08, + "learning_rate": 0.0005, + "loss": 1.0832, + "step": 996 + }, + { + "epoch": 1.08, + "learning_rate": 0.0005, + "loss": 1.0349, + "step": 997 + }, + { + "epoch": 1.09, + "learning_rate": 0.0005, + "loss": 1.1164, + "step": 998 + }, + { + "epoch": 1.09, + "learning_rate": 0.0005, + "loss": 1.0895, + "step": 999 + }, + { + "epoch": 1.09, + "learning_rate": 0.0005, + "loss": 1.0528, + "step": 1000 + }, + { + "epoch": 1.09, + "learning_rate": 0.0005, + "loss": 1.0668, + "step": 1001 + }, + { + "epoch": 1.09, + "learning_rate": 0.0005, + "loss": 1.0905, + "step": 1002 + }, + { + "epoch": 1.09, + "learning_rate": 0.0005, + "loss": 1.0648, + "step": 1003 + }, + { + "epoch": 1.09, + "learning_rate": 0.0005, + "loss": 1.0595, + "step": 1004 + }, + { + "epoch": 1.09, + "learning_rate": 0.0005, + "loss": 1.0799, + "step": 1005 + }, + { + "epoch": 1.09, + "learning_rate": 0.0005, + "loss": 1.1042, + "step": 1006 + }, + { + "epoch": 1.1, + "learning_rate": 0.0005, + "loss": 1.0905, + "step": 1007 + }, + { + "epoch": 1.1, + "learning_rate": 0.0005, + "loss": 1.0995, + "step": 1008 + }, + { + "epoch": 1.1, + "learning_rate": 0.0005, + "loss": 1.0019, + "step": 1009 + }, + { + "epoch": 1.1, + "learning_rate": 0.0005, + "loss": 1.0997, + "step": 1010 + }, + { + "epoch": 1.1, + "learning_rate": 0.0005, + "loss": 1.0654, + "step": 1011 + }, + { + "epoch": 1.1, + "learning_rate": 0.0005, + "loss": 1.0387, + "step": 1012 + }, + { + "epoch": 1.1, + "learning_rate": 0.0005, + "loss": 1.077, + "step": 1013 + }, + { + "epoch": 1.1, + "learning_rate": 0.0005, + "loss": 1.0909, + "step": 1014 + }, + { + "epoch": 1.1, + "learning_rate": 0.0005, + "loss": 1.1072, + "step": 1015 + }, + { + "epoch": 1.11, + "learning_rate": 0.0005, + "loss": 1.0618, + "step": 1016 + }, + { + "epoch": 1.11, + "learning_rate": 0.0005, + "loss": 1.0727, + "step": 1017 + }, + { + "epoch": 1.11, + "learning_rate": 0.0005, + "loss": 1.029, + "step": 1018 + }, + { + "epoch": 1.11, + "learning_rate": 0.0005, + "loss": 1.1166, + "step": 1019 + }, + { + "epoch": 1.11, + "learning_rate": 0.0005, + "loss": 1.0903, + "step": 1020 + }, + { + "epoch": 1.11, + "learning_rate": 0.0005, + "loss": 1.0573, + "step": 1021 + }, + { + "epoch": 1.11, + "learning_rate": 0.0005, + "loss": 1.0717, + "step": 1022 + }, + { + "epoch": 1.11, + "learning_rate": 0.0005, + "loss": 1.0937, + "step": 1023 + }, + { + "epoch": 1.11, + "learning_rate": 0.0005, + "loss": 1.1059, + "step": 1024 + }, + { + "epoch": 1.12, + "learning_rate": 0.0005, + "loss": 1.1148, + "step": 1025 + }, + { + "epoch": 1.12, + "learning_rate": 0.0005, + "loss": 1.0398, + "step": 1026 + }, + { + "epoch": 1.12, + "learning_rate": 0.0005, + "loss": 1.0743, + "step": 1027 + }, + { + "epoch": 1.12, + "learning_rate": 0.0005, + "loss": 1.0518, + "step": 1028 + }, + { + "epoch": 1.12, + "learning_rate": 0.0005, + "loss": 1.0538, + "step": 1029 + }, + { + "epoch": 1.12, + "learning_rate": 0.0005, + "loss": 1.0619, + "step": 1030 + }, + { + "epoch": 1.12, + "learning_rate": 0.0005, + "loss": 1.0694, + "step": 1031 + }, + { + "epoch": 1.12, + "learning_rate": 0.0005, + "loss": 1.0626, + "step": 1032 + }, + { + "epoch": 1.12, + "learning_rate": 0.0005, + "loss": 1.106, + "step": 1033 + }, + { + "epoch": 1.13, + "learning_rate": 0.0005, + "loss": 1.0588, + "step": 1034 + }, + { + "epoch": 1.13, + "learning_rate": 0.0005, + "loss": 1.0473, + "step": 1035 + }, + { + "epoch": 1.13, + "learning_rate": 0.0005, + "loss": 1.0884, + "step": 1036 + }, + { + "epoch": 1.13, + "learning_rate": 0.0005, + "loss": 1.0538, + "step": 1037 + }, + { + "epoch": 1.13, + "learning_rate": 0.0005, + "loss": 1.1239, + "step": 1038 + }, + { + "epoch": 1.13, + "learning_rate": 0.0005, + "loss": 1.0904, + "step": 1039 + }, + { + "epoch": 1.13, + "learning_rate": 0.0005, + "loss": 1.0691, + "step": 1040 + }, + { + "epoch": 1.13, + "learning_rate": 0.0005, + "loss": 1.0954, + "step": 1041 + }, + { + "epoch": 1.13, + "learning_rate": 0.0005, + "loss": 1.0642, + "step": 1042 + }, + { + "epoch": 1.13, + "learning_rate": 0.0005, + "loss": 1.0514, + "step": 1043 + }, + { + "epoch": 1.14, + "learning_rate": 0.0005, + "loss": 1.0748, + "step": 1044 + }, + { + "epoch": 1.14, + "learning_rate": 0.0005, + "loss": 1.0659, + "step": 1045 + }, + { + "epoch": 1.14, + "learning_rate": 0.0005, + "loss": 1.0806, + "step": 1046 + }, + { + "epoch": 1.14, + "learning_rate": 0.0005, + "loss": 1.0548, + "step": 1047 + }, + { + "epoch": 1.14, + "learning_rate": 0.0005, + "loss": 1.0502, + "step": 1048 + }, + { + "epoch": 1.14, + "learning_rate": 0.0005, + "loss": 1.0762, + "step": 1049 + }, + { + "epoch": 1.14, + "learning_rate": 0.0005, + "loss": 1.0604, + "step": 1050 + }, + { + "epoch": 1.14, + "learning_rate": 0.0005, + "loss": 1.1259, + "step": 1051 + }, + { + "epoch": 1.14, + "learning_rate": 0.0005, + "loss": 1.1085, + "step": 1052 + }, + { + "epoch": 1.15, + "learning_rate": 0.0005, + "loss": 1.0541, + "step": 1053 + }, + { + "epoch": 1.15, + "learning_rate": 0.0005, + "loss": 1.1035, + "step": 1054 + }, + { + "epoch": 1.15, + "learning_rate": 0.0005, + "loss": 1.0448, + "step": 1055 + }, + { + "epoch": 1.15, + "learning_rate": 0.0005, + "loss": 1.0902, + "step": 1056 + }, + { + "epoch": 1.15, + "learning_rate": 0.0005, + "loss": 1.1143, + "step": 1057 + }, + { + "epoch": 1.15, + "learning_rate": 0.0005, + "loss": 1.0626, + "step": 1058 + }, + { + "epoch": 1.15, + "learning_rate": 0.0005, + "loss": 1.0604, + "step": 1059 + }, + { + "epoch": 1.15, + "learning_rate": 0.0005, + "loss": 1.0369, + "step": 1060 + }, + { + "epoch": 1.15, + "learning_rate": 0.0005, + "loss": 1.0561, + "step": 1061 + }, + { + "epoch": 1.16, + "learning_rate": 0.0005, + "loss": 1.0506, + "step": 1062 + }, + { + "epoch": 1.16, + "learning_rate": 0.0005, + "loss": 1.0549, + "step": 1063 + }, + { + "epoch": 1.16, + "learning_rate": 0.0005, + "loss": 1.1061, + "step": 1064 + }, + { + "epoch": 1.16, + "learning_rate": 0.0005, + "loss": 1.1122, + "step": 1065 + }, + { + "epoch": 1.16, + "learning_rate": 0.0005, + "loss": 1.0419, + "step": 1066 + }, + { + "epoch": 1.16, + "learning_rate": 0.0005, + "loss": 1.0882, + "step": 1067 + }, + { + "epoch": 1.16, + "learning_rate": 0.0005, + "loss": 1.0764, + "step": 1068 + }, + { + "epoch": 1.16, + "learning_rate": 0.0005, + "loss": 1.1047, + "step": 1069 + }, + { + "epoch": 1.16, + "learning_rate": 0.0005, + "loss": 1.0406, + "step": 1070 + }, + { + "epoch": 1.17, + "learning_rate": 0.0005, + "loss": 1.0623, + "step": 1071 + }, + { + "epoch": 1.17, + "learning_rate": 0.0005, + "loss": 1.0925, + "step": 1072 + }, + { + "epoch": 1.17, + "learning_rate": 0.0005, + "loss": 1.0201, + "step": 1073 + }, + { + "epoch": 1.17, + "learning_rate": 0.0005, + "loss": 1.0886, + "step": 1074 + }, + { + "epoch": 1.17, + "learning_rate": 0.0005, + "loss": 1.0988, + "step": 1075 + }, + { + "epoch": 1.17, + "learning_rate": 0.0005, + "loss": 1.0362, + "step": 1076 + }, + { + "epoch": 1.17, + "learning_rate": 0.0005, + "loss": 1.0624, + "step": 1077 + }, + { + "epoch": 1.17, + "learning_rate": 0.0005, + "loss": 1.0675, + "step": 1078 + }, + { + "epoch": 1.17, + "learning_rate": 0.0005, + "loss": 1.0882, + "step": 1079 + }, + { + "epoch": 1.18, + "learning_rate": 0.0005, + "loss": 1.0132, + "step": 1080 + }, + { + "epoch": 1.18, + "learning_rate": 0.0005, + "loss": 1.0663, + "step": 1081 + }, + { + "epoch": 1.18, + "learning_rate": 0.0005, + "loss": 1.052, + "step": 1082 + }, + { + "epoch": 1.18, + "learning_rate": 0.0005, + "loss": 1.0547, + "step": 1083 + }, + { + "epoch": 1.18, + "learning_rate": 0.0005, + "loss": 1.0034, + "step": 1084 + }, + { + "epoch": 1.18, + "learning_rate": 0.0005, + "loss": 1.0275, + "step": 1085 + }, + { + "epoch": 1.18, + "learning_rate": 0.0005, + "loss": 1.0342, + "step": 1086 + }, + { + "epoch": 1.18, + "learning_rate": 0.0005, + "loss": 1.0777, + "step": 1087 + }, + { + "epoch": 1.18, + "learning_rate": 0.0005, + "loss": 1.0838, + "step": 1088 + }, + { + "epoch": 1.18, + "learning_rate": 0.0005, + "loss": 1.0801, + "step": 1089 + }, + { + "epoch": 1.19, + "learning_rate": 0.0005, + "loss": 1.0613, + "step": 1090 + }, + { + "epoch": 1.19, + "learning_rate": 0.0005, + "loss": 1.0694, + "step": 1091 + }, + { + "epoch": 1.19, + "learning_rate": 0.0005, + "loss": 1.0797, + "step": 1092 + }, + { + "epoch": 1.19, + "learning_rate": 0.0005, + "loss": 1.1009, + "step": 1093 + }, + { + "epoch": 1.19, + "learning_rate": 0.0005, + "loss": 1.0656, + "step": 1094 + }, + { + "epoch": 1.19, + "learning_rate": 0.0005, + "loss": 1.0615, + "step": 1095 + }, + { + "epoch": 1.19, + "learning_rate": 0.0005, + "loss": 1.0727, + "step": 1096 + }, + { + "epoch": 1.19, + "learning_rate": 0.0005, + "loss": 1.0895, + "step": 1097 + }, + { + "epoch": 1.19, + "learning_rate": 0.0005, + "loss": 1.0667, + "step": 1098 + }, + { + "epoch": 1.2, + "learning_rate": 0.0005, + "loss": 1.1004, + "step": 1099 + }, + { + "epoch": 1.2, + "learning_rate": 0.0005, + "loss": 1.0313, + "step": 1100 + }, + { + "epoch": 1.2, + "learning_rate": 0.0005, + "loss": 1.0369, + "step": 1101 + }, + { + "epoch": 1.2, + "learning_rate": 0.0005, + "loss": 1.0662, + "step": 1102 + }, + { + "epoch": 1.2, + "learning_rate": 0.0005, + "loss": 1.0718, + "step": 1103 + }, + { + "epoch": 1.2, + "learning_rate": 0.0005, + "loss": 1.0836, + "step": 1104 + }, + { + "epoch": 1.2, + "learning_rate": 0.0005, + "loss": 1.07, + "step": 1105 + }, + { + "epoch": 1.2, + "learning_rate": 0.0005, + "loss": 1.0382, + "step": 1106 + }, + { + "epoch": 1.2, + "learning_rate": 0.0005, + "loss": 1.0542, + "step": 1107 + }, + { + "epoch": 1.21, + "learning_rate": 0.0005, + "loss": 1.0766, + "step": 1108 + }, + { + "epoch": 1.21, + "learning_rate": 0.0005, + "loss": 1.1121, + "step": 1109 + }, + { + "epoch": 1.21, + "learning_rate": 0.0005, + "loss": 1.0709, + "step": 1110 + }, + { + "epoch": 1.21, + "learning_rate": 0.0005, + "loss": 1.034, + "step": 1111 + }, + { + "epoch": 1.21, + "learning_rate": 0.0005, + "loss": 1.0619, + "step": 1112 + }, + { + "epoch": 1.21, + "learning_rate": 0.0005, + "loss": 1.0739, + "step": 1113 + }, + { + "epoch": 1.21, + "learning_rate": 0.0005, + "loss": 1.1042, + "step": 1114 + }, + { + "epoch": 1.21, + "learning_rate": 0.0005, + "loss": 1.0871, + "step": 1115 + }, + { + "epoch": 1.21, + "learning_rate": 0.0005, + "loss": 1.0933, + "step": 1116 + }, + { + "epoch": 1.22, + "learning_rate": 0.0005, + "loss": 1.0813, + "step": 1117 + }, + { + "epoch": 1.22, + "learning_rate": 0.0005, + "loss": 1.0686, + "step": 1118 + }, + { + "epoch": 1.22, + "learning_rate": 0.0005, + "loss": 1.038, + "step": 1119 + }, + { + "epoch": 1.22, + "learning_rate": 0.0005, + "loss": 1.0284, + "step": 1120 + }, + { + "epoch": 1.22, + "learning_rate": 0.0005, + "loss": 1.0115, + "step": 1121 + }, + { + "epoch": 1.22, + "learning_rate": 0.0005, + "loss": 1.07, + "step": 1122 + }, + { + "epoch": 1.22, + "learning_rate": 0.0005, + "loss": 1.0614, + "step": 1123 + }, + { + "epoch": 1.22, + "learning_rate": 0.0005, + "loss": 1.116, + "step": 1124 + }, + { + "epoch": 1.22, + "learning_rate": 0.0005, + "loss": 1.0447, + "step": 1125 + }, + { + "epoch": 1.23, + "learning_rate": 0.0005, + "loss": 1.0937, + "step": 1126 + }, + { + "epoch": 1.23, + "learning_rate": 0.0005, + "loss": 1.1048, + "step": 1127 + }, + { + "epoch": 1.23, + "learning_rate": 0.0005, + "loss": 1.0691, + "step": 1128 + }, + { + "epoch": 1.23, + "learning_rate": 0.0005, + "loss": 1.0306, + "step": 1129 + }, + { + "epoch": 1.23, + "learning_rate": 0.0005, + "loss": 1.0627, + "step": 1130 + }, + { + "epoch": 1.23, + "learning_rate": 0.0005, + "loss": 1.0736, + "step": 1131 + }, + { + "epoch": 1.23, + "learning_rate": 0.0005, + "loss": 1.05, + "step": 1132 + }, + { + "epoch": 1.23, + "learning_rate": 0.0005, + "loss": 1.0616, + "step": 1133 + }, + { + "epoch": 1.23, + "learning_rate": 0.0005, + "loss": 1.0837, + "step": 1134 + }, + { + "epoch": 1.24, + "learning_rate": 0.0005, + "loss": 1.0988, + "step": 1135 + }, + { + "epoch": 1.24, + "learning_rate": 0.0005, + "loss": 1.0681, + "step": 1136 + }, + { + "epoch": 1.24, + "learning_rate": 0.0005, + "loss": 1.0809, + "step": 1137 + }, + { + "epoch": 1.24, + "learning_rate": 0.0005, + "loss": 1.0394, + "step": 1138 + }, + { + "epoch": 1.24, + "learning_rate": 0.0005, + "loss": 1.0433, + "step": 1139 + }, + { + "epoch": 1.24, + "learning_rate": 0.0005, + "loss": 1.089, + "step": 1140 + }, + { + "epoch": 1.24, + "learning_rate": 0.0005, + "loss": 1.0512, + "step": 1141 + }, + { + "epoch": 1.24, + "learning_rate": 0.0005, + "loss": 1.0503, + "step": 1142 + }, + { + "epoch": 1.24, + "learning_rate": 0.0005, + "loss": 1.0495, + "step": 1143 + }, + { + "epoch": 1.24, + "learning_rate": 0.0005, + "loss": 1.0441, + "step": 1144 + }, + { + "epoch": 1.25, + "learning_rate": 0.0005, + "loss": 1.0973, + "step": 1145 + }, + { + "epoch": 1.25, + "learning_rate": 0.0005, + "loss": 1.0914, + "step": 1146 + }, + { + "epoch": 1.25, + "learning_rate": 0.0005, + "loss": 1.0251, + "step": 1147 + }, + { + "epoch": 1.25, + "learning_rate": 0.0005, + "loss": 1.0892, + "step": 1148 + }, + { + "epoch": 1.25, + "learning_rate": 0.0005, + "loss": 1.1384, + "step": 1149 + }, + { + "epoch": 1.25, + "learning_rate": 0.0005, + "loss": 0.9944, + "step": 1150 + }, + { + "epoch": 1.25, + "learning_rate": 0.0005, + "loss": 1.0652, + "step": 1151 + }, + { + "epoch": 1.25, + "learning_rate": 0.0005, + "loss": 1.0921, + "step": 1152 + }, + { + "epoch": 1.25, + "learning_rate": 0.0005, + "loss": 1.055, + "step": 1153 + }, + { + "epoch": 1.26, + "learning_rate": 0.0005, + "loss": 1.0522, + "step": 1154 + }, + { + "epoch": 1.26, + "learning_rate": 0.0005, + "loss": 1.0991, + "step": 1155 + }, + { + "epoch": 1.26, + "learning_rate": 0.0005, + "loss": 1.0321, + "step": 1156 + }, + { + "epoch": 1.26, + "learning_rate": 0.0005, + "loss": 1.0731, + "step": 1157 + }, + { + "epoch": 1.26, + "learning_rate": 0.0005, + "loss": 1.0658, + "step": 1158 + }, + { + "epoch": 1.26, + "learning_rate": 0.0005, + "loss": 1.0552, + "step": 1159 + }, + { + "epoch": 1.26, + "learning_rate": 0.0005, + "loss": 1.0949, + "step": 1160 + }, + { + "epoch": 1.26, + "learning_rate": 0.0005, + "loss": 1.0318, + "step": 1161 + }, + { + "epoch": 1.26, + "learning_rate": 0.0005, + "loss": 1.0773, + "step": 1162 + }, + { + "epoch": 1.27, + "learning_rate": 0.0005, + "loss": 1.102, + "step": 1163 + }, + { + "epoch": 1.27, + "learning_rate": 0.0005, + "loss": 1.1004, + "step": 1164 + }, + { + "epoch": 1.27, + "learning_rate": 0.0005, + "loss": 0.9993, + "step": 1165 + }, + { + "epoch": 1.27, + "learning_rate": 0.0005, + "loss": 1.1089, + "step": 1166 + }, + { + "epoch": 1.27, + "learning_rate": 0.0005, + "loss": 1.064, + "step": 1167 + }, + { + "epoch": 1.27, + "learning_rate": 0.0005, + "loss": 1.0727, + "step": 1168 + }, + { + "epoch": 1.27, + "learning_rate": 0.0005, + "loss": 1.1036, + "step": 1169 + }, + { + "epoch": 1.27, + "learning_rate": 0.0005, + "loss": 1.0421, + "step": 1170 + }, + { + "epoch": 1.27, + "learning_rate": 0.0005, + "loss": 1.0391, + "step": 1171 + }, + { + "epoch": 1.28, + "learning_rate": 0.0005, + "loss": 1.0921, + "step": 1172 + }, + { + "epoch": 1.28, + "learning_rate": 0.0005, + "loss": 1.0413, + "step": 1173 + }, + { + "epoch": 1.28, + "learning_rate": 0.0005, + "loss": 1.0902, + "step": 1174 + }, + { + "epoch": 1.28, + "learning_rate": 0.0005, + "loss": 1.0401, + "step": 1175 + }, + { + "epoch": 1.28, + "learning_rate": 0.0005, + "loss": 1.0662, + "step": 1176 + }, + { + "epoch": 1.28, + "learning_rate": 0.0005, + "loss": 1.0378, + "step": 1177 + }, + { + "epoch": 1.28, + "learning_rate": 0.0005, + "loss": 1.0191, + "step": 1178 + }, + { + "epoch": 1.28, + "learning_rate": 0.0005, + "loss": 1.073, + "step": 1179 + }, + { + "epoch": 1.28, + "learning_rate": 0.0005, + "loss": 1.032, + "step": 1180 + }, + { + "epoch": 1.29, + "learning_rate": 0.0005, + "loss": 1.066, + "step": 1181 + }, + { + "epoch": 1.29, + "learning_rate": 0.0005, + "loss": 1.0684, + "step": 1182 + }, + { + "epoch": 1.29, + "learning_rate": 0.0005, + "loss": 1.1116, + "step": 1183 + }, + { + "epoch": 1.29, + "learning_rate": 0.0005, + "loss": 1.0759, + "step": 1184 + }, + { + "epoch": 1.29, + "learning_rate": 0.0005, + "loss": 1.0204, + "step": 1185 + }, + { + "epoch": 1.29, + "learning_rate": 0.0005, + "loss": 1.1553, + "step": 1186 + }, + { + "epoch": 1.29, + "learning_rate": 0.0005, + "loss": 1.0519, + "step": 1187 + }, + { + "epoch": 1.29, + "learning_rate": 0.0005, + "loss": 1.0723, + "step": 1188 + }, + { + "epoch": 1.29, + "learning_rate": 0.0005, + "loss": 1.0339, + "step": 1189 + }, + { + "epoch": 1.29, + "learning_rate": 0.0005, + "loss": 1.0694, + "step": 1190 + }, + { + "epoch": 1.3, + "learning_rate": 0.0005, + "loss": 1.0737, + "step": 1191 + }, + { + "epoch": 1.3, + "learning_rate": 0.0005, + "loss": 1.0203, + "step": 1192 + }, + { + "epoch": 1.3, + "learning_rate": 0.0005, + "loss": 1.1175, + "step": 1193 + }, + { + "epoch": 1.3, + "learning_rate": 0.0005, + "loss": 1.0552, + "step": 1194 + }, + { + "epoch": 1.3, + "learning_rate": 0.0005, + "loss": 1.0747, + "step": 1195 + }, + { + "epoch": 1.3, + "learning_rate": 0.0005, + "loss": 1.1005, + "step": 1196 + }, + { + "epoch": 1.3, + "learning_rate": 0.0005, + "loss": 1.1151, + "step": 1197 + }, + { + "epoch": 1.3, + "learning_rate": 0.0005, + "loss": 1.0591, + "step": 1198 + }, + { + "epoch": 1.3, + "learning_rate": 0.0005, + "loss": 1.0696, + "step": 1199 + }, + { + "epoch": 1.31, + "learning_rate": 0.0005, + "loss": 1.0269, + "step": 1200 + }, + { + "epoch": 1.31, + "learning_rate": 0.0005, + "loss": 1.0642, + "step": 1201 + }, + { + "epoch": 1.31, + "learning_rate": 0.0005, + "loss": 1.0688, + "step": 1202 + }, + { + "epoch": 1.31, + "learning_rate": 0.0005, + "loss": 1.0411, + "step": 1203 + }, + { + "epoch": 1.31, + "learning_rate": 0.0005, + "loss": 1.0946, + "step": 1204 + }, + { + "epoch": 1.31, + "learning_rate": 0.0005, + "loss": 1.0759, + "step": 1205 + }, + { + "epoch": 1.31, + "learning_rate": 0.0005, + "loss": 1.076, + "step": 1206 + }, + { + "epoch": 1.31, + "learning_rate": 0.0005, + "loss": 1.0229, + "step": 1207 + }, + { + "epoch": 1.31, + "learning_rate": 0.0005, + "loss": 1.0909, + "step": 1208 + }, + { + "epoch": 1.32, + "learning_rate": 0.0005, + "loss": 1.0242, + "step": 1209 + }, + { + "epoch": 1.32, + "learning_rate": 0.0005, + "loss": 1.082, + "step": 1210 + }, + { + "epoch": 1.32, + "learning_rate": 0.0005, + "loss": 1.0861, + "step": 1211 + }, + { + "epoch": 1.32, + "learning_rate": 0.0005, + "loss": 1.0403, + "step": 1212 + }, + { + "epoch": 1.32, + "learning_rate": 0.0005, + "loss": 1.0747, + "step": 1213 + }, + { + "epoch": 1.32, + "learning_rate": 0.0005, + "loss": 1.0374, + "step": 1214 + }, + { + "epoch": 1.32, + "learning_rate": 0.0005, + "loss": 1.1002, + "step": 1215 + }, + { + "epoch": 1.32, + "learning_rate": 0.0005, + "loss": 1.0097, + "step": 1216 + }, + { + "epoch": 1.32, + "learning_rate": 0.0005, + "loss": 1.046, + "step": 1217 + }, + { + "epoch": 1.33, + "learning_rate": 0.0005, + "loss": 1.0698, + "step": 1218 + }, + { + "epoch": 1.33, + "learning_rate": 0.0005, + "loss": 1.0156, + "step": 1219 + }, + { + "epoch": 1.33, + "learning_rate": 0.0005, + "loss": 1.0907, + "step": 1220 + }, + { + "epoch": 1.33, + "learning_rate": 0.0005, + "loss": 1.0313, + "step": 1221 + }, + { + "epoch": 1.33, + "learning_rate": 0.0005, + "loss": 1.1015, + "step": 1222 + }, + { + "epoch": 1.33, + "learning_rate": 0.0005, + "loss": 1.0869, + "step": 1223 + }, + { + "epoch": 1.33, + "learning_rate": 0.0005, + "loss": 1.0308, + "step": 1224 + }, + { + "epoch": 1.33, + "learning_rate": 0.0005, + "loss": 1.0765, + "step": 1225 + }, + { + "epoch": 1.33, + "learning_rate": 0.0005, + "loss": 1.0537, + "step": 1226 + }, + { + "epoch": 1.34, + "learning_rate": 0.0005, + "loss": 1.0729, + "step": 1227 + }, + { + "epoch": 1.34, + "learning_rate": 0.0005, + "loss": 1.0896, + "step": 1228 + }, + { + "epoch": 1.34, + "learning_rate": 0.0005, + "loss": 1.0401, + "step": 1229 + }, + { + "epoch": 1.34, + "learning_rate": 0.0005, + "loss": 1.0908, + "step": 1230 + }, + { + "epoch": 1.34, + "learning_rate": 0.0005, + "loss": 1.0926, + "step": 1231 + }, + { + "epoch": 1.34, + "learning_rate": 0.0005, + "loss": 1.0808, + "step": 1232 + }, + { + "epoch": 1.34, + "learning_rate": 0.0005, + "loss": 1.0669, + "step": 1233 + }, + { + "epoch": 1.34, + "learning_rate": 0.0005, + "loss": 1.0996, + "step": 1234 + }, + { + "epoch": 1.34, + "learning_rate": 0.0005, + "loss": 1.1091, + "step": 1235 + }, + { + "epoch": 1.34, + "learning_rate": 0.0005, + "loss": 1.0666, + "step": 1236 + }, + { + "epoch": 1.35, + "learning_rate": 0.0005, + "loss": 1.0298, + "step": 1237 + }, + { + "epoch": 1.35, + "learning_rate": 0.0005, + "loss": 1.0701, + "step": 1238 + }, + { + "epoch": 1.35, + "learning_rate": 0.0005, + "loss": 1.0959, + "step": 1239 + }, + { + "epoch": 1.35, + "learning_rate": 0.0005, + "loss": 1.0953, + "step": 1240 + }, + { + "epoch": 1.35, + "learning_rate": 0.0005, + "loss": 1.0535, + "step": 1241 + }, + { + "epoch": 1.35, + "learning_rate": 0.0005, + "loss": 1.0941, + "step": 1242 + }, + { + "epoch": 1.35, + "learning_rate": 0.0005, + "loss": 1.0767, + "step": 1243 + }, + { + "epoch": 1.35, + "learning_rate": 0.0005, + "loss": 1.0492, + "step": 1244 + }, + { + "epoch": 1.35, + "learning_rate": 0.0005, + "loss": 1.039, + "step": 1245 + }, + { + "epoch": 1.36, + "learning_rate": 0.0005, + "loss": 1.1024, + "step": 1246 + }, + { + "epoch": 1.36, + "learning_rate": 0.0005, + "loss": 1.0582, + "step": 1247 + }, + { + "epoch": 1.36, + "learning_rate": 0.0005, + "loss": 1.0738, + "step": 1248 + }, + { + "epoch": 1.36, + "learning_rate": 0.0005, + "loss": 1.1286, + "step": 1249 + }, + { + "epoch": 1.36, + "learning_rate": 0.0005, + "loss": 1.0303, + "step": 1250 + }, + { + "epoch": 1.36, + "learning_rate": 0.0005, + "loss": 1.0776, + "step": 1251 + }, + { + "epoch": 1.36, + "learning_rate": 0.0005, + "loss": 1.0962, + "step": 1252 + }, + { + "epoch": 1.36, + "learning_rate": 0.0005, + "loss": 1.0763, + "step": 1253 + }, + { + "epoch": 1.36, + "learning_rate": 0.0005, + "loss": 1.0467, + "step": 1254 + }, + { + "epoch": 1.37, + "learning_rate": 0.0005, + "loss": 1.1063, + "step": 1255 + }, + { + "epoch": 1.37, + "learning_rate": 0.0005, + "loss": 1.0362, + "step": 1256 + }, + { + "epoch": 1.37, + "learning_rate": 0.0005, + "loss": 1.0789, + "step": 1257 + }, + { + "epoch": 1.37, + "learning_rate": 0.0005, + "loss": 1.0816, + "step": 1258 + }, + { + "epoch": 1.37, + "learning_rate": 0.0005, + "loss": 1.0791, + "step": 1259 + }, + { + "epoch": 1.37, + "learning_rate": 0.0005, + "loss": 1.0435, + "step": 1260 + }, + { + "epoch": 1.37, + "learning_rate": 0.0005, + "loss": 1.0586, + "step": 1261 + }, + { + "epoch": 1.37, + "learning_rate": 0.0005, + "loss": 1.0366, + "step": 1262 + }, + { + "epoch": 1.37, + "learning_rate": 0.0005, + "loss": 1.0782, + "step": 1263 + }, + { + "epoch": 1.38, + "learning_rate": 0.0005, + "loss": 1.0145, + "step": 1264 + }, + { + "epoch": 1.38, + "learning_rate": 0.0005, + "loss": 1.0838, + "step": 1265 + }, + { + "epoch": 1.38, + "learning_rate": 0.0005, + "loss": 1.082, + "step": 1266 + }, + { + "epoch": 1.38, + "learning_rate": 0.0005, + "loss": 1.0848, + "step": 1267 + }, + { + "epoch": 1.38, + "learning_rate": 0.0005, + "loss": 0.9894, + "step": 1268 + }, + { + "epoch": 1.38, + "learning_rate": 0.0005, + "loss": 1.05, + "step": 1269 + }, + { + "epoch": 1.38, + "learning_rate": 0.0005, + "loss": 1.0588, + "step": 1270 + }, + { + "epoch": 1.38, + "learning_rate": 0.0005, + "loss": 1.0921, + "step": 1271 + }, + { + "epoch": 1.38, + "learning_rate": 0.0005, + "loss": 1.0498, + "step": 1272 + }, + { + "epoch": 1.39, + "learning_rate": 0.0005, + "loss": 1.0644, + "step": 1273 + }, + { + "epoch": 1.39, + "learning_rate": 0.0005, + "loss": 1.1227, + "step": 1274 + }, + { + "epoch": 1.39, + "learning_rate": 0.0005, + "loss": 1.1029, + "step": 1275 + }, + { + "epoch": 1.39, + "learning_rate": 0.0005, + "loss": 1.0895, + "step": 1276 + }, + { + "epoch": 1.39, + "learning_rate": 0.0005, + "loss": 1.0725, + "step": 1277 + }, + { + "epoch": 1.39, + "learning_rate": 0.0005, + "loss": 1.0312, + "step": 1278 + }, + { + "epoch": 1.39, + "learning_rate": 0.0005, + "loss": 1.0054, + "step": 1279 + }, + { + "epoch": 1.39, + "learning_rate": 0.0005, + "loss": 1.0288, + "step": 1280 + }, + { + "epoch": 1.39, + "learning_rate": 0.0005, + "loss": 1.0638, + "step": 1281 + }, + { + "epoch": 1.39, + "learning_rate": 0.0005, + "loss": 1.0689, + "step": 1282 + }, + { + "epoch": 1.4, + "learning_rate": 0.0005, + "loss": 1.0527, + "step": 1283 + }, + { + "epoch": 1.4, + "learning_rate": 0.0005, + "loss": 1.0799, + "step": 1284 + }, + { + "epoch": 1.4, + "learning_rate": 0.0005, + "loss": 1.0548, + "step": 1285 + }, + { + "epoch": 1.4, + "learning_rate": 0.0005, + "loss": 1.0705, + "step": 1286 + }, + { + "epoch": 1.4, + "learning_rate": 0.0005, + "loss": 1.0772, + "step": 1287 + }, + { + "epoch": 1.4, + "learning_rate": 0.0005, + "loss": 1.0712, + "step": 1288 + }, + { + "epoch": 1.4, + "learning_rate": 0.0005, + "loss": 1.0704, + "step": 1289 + }, + { + "epoch": 1.4, + "learning_rate": 0.0005, + "loss": 1.0562, + "step": 1290 + }, + { + "epoch": 1.4, + "learning_rate": 0.0005, + "loss": 1.0494, + "step": 1291 + }, + { + "epoch": 1.41, + "learning_rate": 0.0005, + "loss": 1.0542, + "step": 1292 + }, + { + "epoch": 1.41, + "learning_rate": 0.0005, + "loss": 1.0769, + "step": 1293 + }, + { + "epoch": 1.41, + "learning_rate": 0.0005, + "loss": 1.0622, + "step": 1294 + }, + { + "epoch": 1.41, + "learning_rate": 0.0005, + "loss": 1.0196, + "step": 1295 + }, + { + "epoch": 1.41, + "learning_rate": 0.0005, + "loss": 1.0825, + "step": 1296 + }, + { + "epoch": 1.41, + "learning_rate": 0.0005, + "loss": 1.0963, + "step": 1297 + }, + { + "epoch": 1.41, + "learning_rate": 0.0005, + "loss": 1.0795, + "step": 1298 + }, + { + "epoch": 1.41, + "learning_rate": 0.0005, + "loss": 1.0547, + "step": 1299 + }, + { + "epoch": 1.41, + "learning_rate": 0.0005, + "loss": 1.0682, + "step": 1300 + }, + { + "epoch": 1.42, + "learning_rate": 0.0005, + "loss": 1.0582, + "step": 1301 + }, + { + "epoch": 1.42, + "learning_rate": 0.0005, + "loss": 1.0899, + "step": 1302 + }, + { + "epoch": 1.42, + "learning_rate": 0.0005, + "loss": 1.018, + "step": 1303 + }, + { + "epoch": 1.42, + "learning_rate": 0.0005, + "loss": 1.0727, + "step": 1304 + }, + { + "epoch": 1.42, + "learning_rate": 0.0005, + "loss": 1.0654, + "step": 1305 + }, + { + "epoch": 1.42, + "learning_rate": 0.0005, + "loss": 1.0822, + "step": 1306 + }, + { + "epoch": 1.42, + "learning_rate": 0.0005, + "loss": 1.0513, + "step": 1307 + }, + { + "epoch": 1.42, + "learning_rate": 0.0005, + "loss": 1.095, + "step": 1308 + }, + { + "epoch": 1.42, + "learning_rate": 0.0005, + "loss": 1.0799, + "step": 1309 + }, + { + "epoch": 1.43, + "learning_rate": 0.0005, + "loss": 1.0422, + "step": 1310 + }, + { + "epoch": 1.43, + "learning_rate": 0.0005, + "loss": 1.088, + "step": 1311 + }, + { + "epoch": 1.43, + "learning_rate": 0.0005, + "loss": 1.0544, + "step": 1312 + }, + { + "epoch": 1.43, + "learning_rate": 0.0005, + "loss": 1.0794, + "step": 1313 + }, + { + "epoch": 1.43, + "learning_rate": 0.0005, + "loss": 1.0539, + "step": 1314 + }, + { + "epoch": 1.43, + "learning_rate": 0.0005, + "loss": 1.0558, + "step": 1315 + }, + { + "epoch": 1.43, + "learning_rate": 0.0005, + "loss": 1.0247, + "step": 1316 + }, + { + "epoch": 1.43, + "learning_rate": 0.0005, + "loss": 1.0634, + "step": 1317 + }, + { + "epoch": 1.43, + "learning_rate": 0.0005, + "loss": 1.0458, + "step": 1318 + }, + { + "epoch": 1.44, + "learning_rate": 0.0005, + "loss": 1.1118, + "step": 1319 + }, + { + "epoch": 1.44, + "learning_rate": 0.0005, + "loss": 1.0631, + "step": 1320 + }, + { + "epoch": 1.44, + "learning_rate": 0.0005, + "loss": 1.0569, + "step": 1321 + }, + { + "epoch": 1.44, + "learning_rate": 0.0005, + "loss": 1.0375, + "step": 1322 + }, + { + "epoch": 1.44, + "learning_rate": 0.0005, + "loss": 1.0205, + "step": 1323 + }, + { + "epoch": 1.44, + "learning_rate": 0.0005, + "loss": 1.0842, + "step": 1324 + }, + { + "epoch": 1.44, + "learning_rate": 0.0005, + "loss": 1.0998, + "step": 1325 + }, + { + "epoch": 1.44, + "learning_rate": 0.0005, + "loss": 1.1207, + "step": 1326 + }, + { + "epoch": 1.44, + "learning_rate": 0.0005, + "loss": 1.1173, + "step": 1327 + }, + { + "epoch": 1.45, + "learning_rate": 0.0005, + "loss": 1.0473, + "step": 1328 + }, + { + "epoch": 1.45, + "learning_rate": 0.0005, + "loss": 1.043, + "step": 1329 + }, + { + "epoch": 1.45, + "learning_rate": 0.0005, + "loss": 1.0763, + "step": 1330 + }, + { + "epoch": 1.45, + "learning_rate": 0.0005, + "loss": 1.0604, + "step": 1331 + }, + { + "epoch": 1.45, + "learning_rate": 0.0005, + "loss": 1.0215, + "step": 1332 + }, + { + "epoch": 1.45, + "learning_rate": 0.0005, + "loss": 1.0175, + "step": 1333 + }, + { + "epoch": 1.45, + "learning_rate": 0.0005, + "loss": 1.0788, + "step": 1334 + }, + { + "epoch": 1.45, + "learning_rate": 0.0005, + "loss": 1.0763, + "step": 1335 + }, + { + "epoch": 1.45, + "learning_rate": 0.0005, + "loss": 1.0189, + "step": 1336 + }, + { + "epoch": 1.45, + "learning_rate": 0.0005, + "loss": 1.1015, + "step": 1337 + }, + { + "epoch": 1.46, + "learning_rate": 0.0005, + "loss": 1.1427, + "step": 1338 + }, + { + "epoch": 1.46, + "learning_rate": 0.0005, + "loss": 1.0598, + "step": 1339 + }, + { + "epoch": 1.46, + "learning_rate": 0.0005, + "loss": 1.0746, + "step": 1340 + }, + { + "epoch": 1.46, + "learning_rate": 0.0005, + "loss": 1.0563, + "step": 1341 + }, + { + "epoch": 1.46, + "learning_rate": 0.0005, + "loss": 1.0833, + "step": 1342 + }, + { + "epoch": 1.46, + "learning_rate": 0.0005, + "loss": 0.9852, + "step": 1343 + }, + { + "epoch": 1.46, + "learning_rate": 0.0005, + "loss": 1.0212, + "step": 1344 + }, + { + "epoch": 1.46, + "learning_rate": 0.0005, + "loss": 1.0358, + "step": 1345 + }, + { + "epoch": 1.46, + "learning_rate": 0.0005, + "loss": 1.0762, + "step": 1346 + }, + { + "epoch": 1.47, + "learning_rate": 0.0005, + "loss": 1.0636, + "step": 1347 + }, + { + "epoch": 1.47, + "learning_rate": 0.0005, + "loss": 1.0908, + "step": 1348 + }, + { + "epoch": 1.47, + "learning_rate": 0.0005, + "loss": 1.0893, + "step": 1349 + }, + { + "epoch": 1.47, + "learning_rate": 0.0005, + "loss": 1.0481, + "step": 1350 + }, + { + "epoch": 1.47, + "learning_rate": 0.0005, + "loss": 1.1203, + "step": 1351 + }, + { + "epoch": 1.47, + "learning_rate": 0.0005, + "loss": 1.1084, + "step": 1352 + }, + { + "epoch": 1.47, + "learning_rate": 0.0005, + "loss": 1.0815, + "step": 1353 + }, + { + "epoch": 1.47, + "learning_rate": 0.0005, + "loss": 1.086, + "step": 1354 + }, + { + "epoch": 1.47, + "learning_rate": 0.0005, + "loss": 1.0246, + "step": 1355 + }, + { + "epoch": 1.48, + "learning_rate": 0.0005, + "loss": 1.0751, + "step": 1356 + }, + { + "epoch": 1.48, + "learning_rate": 0.0005, + "loss": 1.1185, + "step": 1357 + }, + { + "epoch": 1.48, + "learning_rate": 0.0005, + "loss": 1.0735, + "step": 1358 + }, + { + "epoch": 1.48, + "learning_rate": 0.0005, + "loss": 1.082, + "step": 1359 + }, + { + "epoch": 1.48, + "learning_rate": 0.0005, + "loss": 1.0527, + "step": 1360 + }, + { + "epoch": 1.48, + "learning_rate": 0.0005, + "loss": 1.0906, + "step": 1361 + }, + { + "epoch": 1.48, + "learning_rate": 0.0005, + "loss": 1.093, + "step": 1362 + }, + { + "epoch": 1.48, + "learning_rate": 0.0005, + "loss": 1.0733, + "step": 1363 + }, + { + "epoch": 1.48, + "learning_rate": 0.0005, + "loss": 1.1012, + "step": 1364 + }, + { + "epoch": 1.49, + "learning_rate": 0.0005, + "loss": 1.1038, + "step": 1365 + }, + { + "epoch": 1.49, + "learning_rate": 0.0005, + "loss": 1.0263, + "step": 1366 + }, + { + "epoch": 1.49, + "learning_rate": 0.0005, + "loss": 1.058, + "step": 1367 + }, + { + "epoch": 1.49, + "learning_rate": 0.0005, + "loss": 1.1046, + "step": 1368 + }, + { + "epoch": 1.49, + "learning_rate": 0.0005, + "loss": 1.0256, + "step": 1369 + }, + { + "epoch": 1.49, + "learning_rate": 0.0005, + "loss": 1.0759, + "step": 1370 + }, + { + "epoch": 1.49, + "learning_rate": 0.0005, + "loss": 1.0633, + "step": 1371 + }, + { + "epoch": 1.49, + "learning_rate": 0.0005, + "loss": 1.0734, + "step": 1372 + }, + { + "epoch": 1.49, + "learning_rate": 0.0005, + "loss": 1.0925, + "step": 1373 + }, + { + "epoch": 1.5, + "learning_rate": 0.0005, + "loss": 1.0471, + "step": 1374 + }, + { + "epoch": 1.5, + "learning_rate": 0.0005, + "loss": 1.0305, + "step": 1375 + }, + { + "epoch": 1.5, + "learning_rate": 0.0005, + "loss": 1.0848, + "step": 1376 + }, + { + "epoch": 1.5, + "learning_rate": 0.0005, + "loss": 1.0674, + "step": 1377 + }, + { + "epoch": 1.5, + "learning_rate": 0.0005, + "loss": 1.0519, + "step": 1378 + }, + { + "epoch": 1.5, + "learning_rate": 0.0005, + "loss": 1.1066, + "step": 1379 + }, + { + "epoch": 1.5, + "learning_rate": 0.0005, + "loss": 1.078, + "step": 1380 + }, + { + "epoch": 1.5, + "learning_rate": 0.0005, + "loss": 1.0562, + "step": 1381 + }, + { + "epoch": 1.5, + "learning_rate": 0.0005, + "loss": 1.0809, + "step": 1382 + }, + { + "epoch": 1.5, + "learning_rate": 0.0005, + "loss": 1.0432, + "step": 1383 + }, + { + "epoch": 1.51, + "learning_rate": 0.0005, + "loss": 1.0717, + "step": 1384 + }, + { + "epoch": 1.51, + "learning_rate": 0.0005, + "loss": 1.1016, + "step": 1385 + }, + { + "epoch": 1.51, + "learning_rate": 0.0005, + "loss": 1.0948, + "step": 1386 + }, + { + "epoch": 1.51, + "learning_rate": 0.0005, + "loss": 1.0501, + "step": 1387 + }, + { + "epoch": 1.51, + "learning_rate": 0.0005, + "loss": 1.0555, + "step": 1388 + }, + { + "epoch": 1.51, + "learning_rate": 0.0005, + "loss": 1.0904, + "step": 1389 + }, + { + "epoch": 1.51, + "learning_rate": 0.0005, + "loss": 1.0913, + "step": 1390 + }, + { + "epoch": 1.51, + "learning_rate": 0.0005, + "loss": 1.0423, + "step": 1391 + }, + { + "epoch": 1.51, + "learning_rate": 0.0005, + "loss": 1.1014, + "step": 1392 + }, + { + "epoch": 1.52, + "learning_rate": 0.0005, + "loss": 1.047, + "step": 1393 + }, + { + "epoch": 1.52, + "learning_rate": 0.0005, + "loss": 1.1214, + "step": 1394 + }, + { + "epoch": 1.52, + "learning_rate": 0.0005, + "loss": 1.0324, + "step": 1395 + }, + { + "epoch": 1.52, + "learning_rate": 0.0005, + "loss": 1.0591, + "step": 1396 + }, + { + "epoch": 1.52, + "learning_rate": 0.0005, + "loss": 1.0336, + "step": 1397 + }, + { + "epoch": 1.52, + "learning_rate": 0.0005, + "loss": 1.0546, + "step": 1398 + }, + { + "epoch": 1.52, + "learning_rate": 0.0005, + "loss": 1.0081, + "step": 1399 + }, + { + "epoch": 1.52, + "learning_rate": 0.0005, + "loss": 1.0221, + "step": 1400 + }, + { + "epoch": 1.52, + "learning_rate": 0.0005, + "loss": 1.0844, + "step": 1401 + }, + { + "epoch": 1.53, + "learning_rate": 0.0005, + "loss": 1.058, + "step": 1402 + }, + { + "epoch": 1.53, + "learning_rate": 0.0005, + "loss": 1.0204, + "step": 1403 + }, + { + "epoch": 1.53, + "learning_rate": 0.0005, + "loss": 1.0326, + "step": 1404 + }, + { + "epoch": 1.53, + "learning_rate": 0.0005, + "loss": 1.0574, + "step": 1405 + }, + { + "epoch": 1.53, + "learning_rate": 0.0005, + "loss": 1.0596, + "step": 1406 + }, + { + "epoch": 1.53, + "learning_rate": 0.0005, + "loss": 1.0437, + "step": 1407 + }, + { + "epoch": 1.53, + "learning_rate": 0.0005, + "loss": 1.047, + "step": 1408 + }, + { + "epoch": 1.53, + "learning_rate": 0.0005, + "loss": 1.0631, + "step": 1409 + }, + { + "epoch": 1.53, + "learning_rate": 0.0005, + "loss": 1.0771, + "step": 1410 + }, + { + "epoch": 1.54, + "learning_rate": 0.0005, + "loss": 1.0518, + "step": 1411 + }, + { + "epoch": 1.54, + "learning_rate": 0.0005, + "loss": 1.0826, + "step": 1412 + }, + { + "epoch": 1.54, + "learning_rate": 0.0005, + "loss": 1.0493, + "step": 1413 + }, + { + "epoch": 1.54, + "learning_rate": 0.0005, + "loss": 1.0062, + "step": 1414 + }, + { + "epoch": 1.54, + "learning_rate": 0.0005, + "loss": 1.0739, + "step": 1415 + }, + { + "epoch": 1.54, + "learning_rate": 0.0005, + "loss": 1.0457, + "step": 1416 + }, + { + "epoch": 1.54, + "learning_rate": 0.0005, + "loss": 1.0537, + "step": 1417 + }, + { + "epoch": 1.54, + "learning_rate": 0.0005, + "loss": 1.0754, + "step": 1418 + }, + { + "epoch": 1.54, + "learning_rate": 0.0005, + "loss": 1.1068, + "step": 1419 + }, + { + "epoch": 1.55, + "learning_rate": 0.0005, + "loss": 1.0845, + "step": 1420 + }, + { + "epoch": 1.55, + "learning_rate": 0.0005, + "loss": 1.0656, + "step": 1421 + }, + { + "epoch": 1.55, + "learning_rate": 0.0005, + "loss": 1.0262, + "step": 1422 + }, + { + "epoch": 1.55, + "learning_rate": 0.0005, + "loss": 1.0843, + "step": 1423 + }, + { + "epoch": 1.55, + "learning_rate": 0.0005, + "loss": 1.0418, + "step": 1424 + }, + { + "epoch": 1.55, + "learning_rate": 0.0005, + "loss": 1.0305, + "step": 1425 + }, + { + "epoch": 1.55, + "learning_rate": 0.0005, + "loss": 1.0705, + "step": 1426 + }, + { + "epoch": 1.55, + "learning_rate": 0.0005, + "loss": 1.0406, + "step": 1427 + }, + { + "epoch": 1.55, + "learning_rate": 0.0005, + "loss": 1.0391, + "step": 1428 + }, + { + "epoch": 1.55, + "learning_rate": 0.0005, + "loss": 1.0179, + "step": 1429 + }, + { + "epoch": 1.56, + "learning_rate": 0.0005, + "loss": 1.04, + "step": 1430 + }, + { + "epoch": 1.56, + "learning_rate": 0.0005, + "loss": 1.0642, + "step": 1431 + }, + { + "epoch": 1.56, + "learning_rate": 0.0005, + "loss": 1.0683, + "step": 1432 + }, + { + "epoch": 1.56, + "learning_rate": 0.0005, + "loss": 1.0808, + "step": 1433 + }, + { + "epoch": 1.56, + "learning_rate": 0.0005, + "loss": 1.075, + "step": 1434 + }, + { + "epoch": 1.56, + "learning_rate": 0.0005, + "loss": 1.1017, + "step": 1435 + }, + { + "epoch": 1.56, + "learning_rate": 0.0005, + "loss": 1.0561, + "step": 1436 + }, + { + "epoch": 1.56, + "learning_rate": 0.0005, + "loss": 1.0467, + "step": 1437 + }, + { + "epoch": 1.56, + "learning_rate": 0.0005, + "loss": 1.0786, + "step": 1438 + }, + { + "epoch": 1.57, + "learning_rate": 0.0005, + "loss": 1.0325, + "step": 1439 + }, + { + "epoch": 1.57, + "learning_rate": 0.0005, + "loss": 1.0951, + "step": 1440 + }, + { + "epoch": 1.57, + "learning_rate": 0.0005, + "loss": 1.0151, + "step": 1441 + }, + { + "epoch": 1.57, + "learning_rate": 0.0005, + "loss": 1.0678, + "step": 1442 + }, + { + "epoch": 1.57, + "learning_rate": 0.0005, + "loss": 1.0763, + "step": 1443 + }, + { + "epoch": 1.57, + "learning_rate": 0.0005, + "loss": 1.0255, + "step": 1444 + }, + { + "epoch": 1.57, + "learning_rate": 0.0005, + "loss": 0.9733, + "step": 1445 + }, + { + "epoch": 1.57, + "learning_rate": 0.0005, + "loss": 1.0754, + "step": 1446 + }, + { + "epoch": 1.57, + "learning_rate": 0.0005, + "loss": 1.0805, + "step": 1447 + }, + { + "epoch": 1.58, + "learning_rate": 0.0005, + "loss": 1.0553, + "step": 1448 + }, + { + "epoch": 1.58, + "learning_rate": 0.0005, + "loss": 1.0491, + "step": 1449 + }, + { + "epoch": 1.58, + "learning_rate": 0.0005, + "loss": 1.0502, + "step": 1450 + }, + { + "epoch": 1.58, + "learning_rate": 0.0005, + "loss": 1.0445, + "step": 1451 + }, + { + "epoch": 1.58, + "learning_rate": 0.0005, + "loss": 1.0707, + "step": 1452 + }, + { + "epoch": 1.58, + "learning_rate": 0.0005, + "loss": 1.0556, + "step": 1453 + }, + { + "epoch": 1.58, + "learning_rate": 0.0005, + "loss": 1.0541, + "step": 1454 + }, + { + "epoch": 1.58, + "learning_rate": 0.0005, + "loss": 1.0496, + "step": 1455 + }, + { + "epoch": 1.58, + "learning_rate": 0.0005, + "loss": 1.0417, + "step": 1456 + }, + { + "epoch": 1.59, + "learning_rate": 0.0005, + "loss": 1.0127, + "step": 1457 + }, + { + "epoch": 1.59, + "learning_rate": 0.0005, + "loss": 1.0555, + "step": 1458 + }, + { + "epoch": 1.59, + "learning_rate": 0.0005, + "loss": 1.043, + "step": 1459 + }, + { + "epoch": 1.59, + "learning_rate": 0.0005, + "loss": 1.0479, + "step": 1460 + }, + { + "epoch": 1.59, + "learning_rate": 0.0005, + "loss": 1.0919, + "step": 1461 + }, + { + "epoch": 1.59, + "learning_rate": 0.0005, + "loss": 1.0504, + "step": 1462 + }, + { + "epoch": 1.59, + "learning_rate": 0.0005, + "loss": 1.0543, + "step": 1463 + }, + { + "epoch": 1.59, + "learning_rate": 0.0005, + "loss": 1.0579, + "step": 1464 + }, + { + "epoch": 1.59, + "learning_rate": 0.0005, + "loss": 1.0748, + "step": 1465 + }, + { + "epoch": 1.6, + "learning_rate": 0.0005, + "loss": 1.0682, + "step": 1466 + }, + { + "epoch": 1.6, + "learning_rate": 0.0005, + "loss": 1.0467, + "step": 1467 + }, + { + "epoch": 1.6, + "learning_rate": 0.0005, + "loss": 1.0654, + "step": 1468 + }, + { + "epoch": 1.6, + "learning_rate": 0.0005, + "loss": 1.0508, + "step": 1469 + }, + { + "epoch": 1.6, + "learning_rate": 0.0005, + "loss": 1.0327, + "step": 1470 + }, + { + "epoch": 1.6, + "learning_rate": 0.0005, + "loss": 1.0891, + "step": 1471 + }, + { + "epoch": 1.6, + "learning_rate": 0.0005, + "loss": 1.0454, + "step": 1472 + }, + { + "epoch": 1.6, + "learning_rate": 0.0005, + "loss": 1.1117, + "step": 1473 + }, + { + "epoch": 1.6, + "learning_rate": 0.0005, + "loss": 1.0609, + "step": 1474 + }, + { + "epoch": 1.61, + "learning_rate": 0.0005, + "loss": 1.0514, + "step": 1475 + }, + { + "epoch": 1.61, + "learning_rate": 0.0005, + "loss": 1.1065, + "step": 1476 + }, + { + "epoch": 1.61, + "learning_rate": 0.0005, + "loss": 1.0525, + "step": 1477 + }, + { + "epoch": 1.61, + "learning_rate": 0.0005, + "loss": 1.0709, + "step": 1478 + }, + { + "epoch": 1.61, + "learning_rate": 0.0005, + "loss": 1.0768, + "step": 1479 + }, + { + "epoch": 1.61, + "learning_rate": 0.0005, + "loss": 1.0079, + "step": 1480 + }, + { + "epoch": 1.61, + "learning_rate": 0.0005, + "loss": 1.0598, + "step": 1481 + }, + { + "epoch": 1.61, + "learning_rate": 0.0005, + "loss": 1.0898, + "step": 1482 + }, + { + "epoch": 1.61, + "learning_rate": 0.0005, + "loss": 1.0774, + "step": 1483 + }, + { + "epoch": 1.61, + "learning_rate": 0.0005, + "loss": 1.0408, + "step": 1484 + }, + { + "epoch": 1.62, + "learning_rate": 0.0005, + "loss": 1.0334, + "step": 1485 + }, + { + "epoch": 1.62, + "learning_rate": 0.0005, + "loss": 1.0302, + "step": 1486 + }, + { + "epoch": 1.62, + "learning_rate": 0.0005, + "loss": 1.0482, + "step": 1487 + }, + { + "epoch": 1.62, + "learning_rate": 0.0005, + "loss": 1.0677, + "step": 1488 + }, + { + "epoch": 1.62, + "learning_rate": 0.0005, + "loss": 1.0425, + "step": 1489 + }, + { + "epoch": 1.62, + "learning_rate": 0.0005, + "loss": 1.0428, + "step": 1490 + }, + { + "epoch": 1.62, + "learning_rate": 0.0005, + "loss": 1.0403, + "step": 1491 + }, + { + "epoch": 1.62, + "learning_rate": 0.0005, + "loss": 1.0323, + "step": 1492 + }, + { + "epoch": 1.62, + "learning_rate": 0.0005, + "loss": 1.0767, + "step": 1493 + }, + { + "epoch": 1.63, + "learning_rate": 0.0005, + "loss": 1.046, + "step": 1494 + }, + { + "epoch": 1.63, + "learning_rate": 0.0005, + "loss": 1.0459, + "step": 1495 + }, + { + "epoch": 1.63, + "learning_rate": 0.0005, + "loss": 1.0638, + "step": 1496 + }, + { + "epoch": 1.63, + "learning_rate": 0.0005, + "loss": 1.0548, + "step": 1497 + }, + { + "epoch": 1.63, + "learning_rate": 0.0005, + "loss": 1.0506, + "step": 1498 + }, + { + "epoch": 1.63, + "learning_rate": 0.0005, + "loss": 1.1086, + "step": 1499 + }, + { + "epoch": 1.63, + "learning_rate": 0.0005, + "loss": 1.0579, + "step": 1500 + }, + { + "epoch": 1.63, + "learning_rate": 0.0005, + "loss": 1.053, + "step": 1501 + }, + { + "epoch": 1.63, + "learning_rate": 0.0005, + "loss": 1.0686, + "step": 1502 + }, + { + "epoch": 1.64, + "learning_rate": 0.0005, + "loss": 1.0589, + "step": 1503 + }, + { + "epoch": 1.64, + "learning_rate": 0.0005, + "loss": 1.0708, + "step": 1504 + }, + { + "epoch": 1.64, + "learning_rate": 0.0005, + "loss": 1.041, + "step": 1505 + }, + { + "epoch": 1.64, + "learning_rate": 0.0005, + "loss": 1.0597, + "step": 1506 + }, + { + "epoch": 1.64, + "learning_rate": 0.0005, + "loss": 1.0296, + "step": 1507 + }, + { + "epoch": 1.64, + "learning_rate": 0.0005, + "loss": 1.0931, + "step": 1508 + }, + { + "epoch": 1.64, + "learning_rate": 0.0005, + "loss": 1.0316, + "step": 1509 + }, + { + "epoch": 1.64, + "learning_rate": 0.0005, + "loss": 1.0524, + "step": 1510 + }, + { + "epoch": 1.64, + "learning_rate": 0.0005, + "loss": 1.0924, + "step": 1511 + }, + { + "epoch": 1.65, + "learning_rate": 0.0005, + "loss": 1.1181, + "step": 1512 + }, + { + "epoch": 1.65, + "learning_rate": 0.0005, + "loss": 1.0607, + "step": 1513 + }, + { + "epoch": 1.65, + "learning_rate": 0.0005, + "loss": 1.0453, + "step": 1514 + }, + { + "epoch": 1.65, + "learning_rate": 0.0005, + "loss": 1.0691, + "step": 1515 + }, + { + "epoch": 1.65, + "learning_rate": 0.0005, + "loss": 1.0818, + "step": 1516 + }, + { + "epoch": 1.65, + "learning_rate": 0.0005, + "loss": 1.0534, + "step": 1517 + }, + { + "epoch": 1.65, + "learning_rate": 0.0005, + "loss": 1.0226, + "step": 1518 + }, + { + "epoch": 1.65, + "learning_rate": 0.0005, + "loss": 1.1018, + "step": 1519 + }, + { + "epoch": 1.65, + "learning_rate": 0.0005, + "loss": 1.0301, + "step": 1520 + }, + { + "epoch": 1.66, + "learning_rate": 0.0005, + "loss": 1.0705, + "step": 1521 + }, + { + "epoch": 1.66, + "learning_rate": 0.0005, + "loss": 1.0603, + "step": 1522 + }, + { + "epoch": 1.66, + "learning_rate": 0.0005, + "loss": 1.0427, + "step": 1523 + }, + { + "epoch": 1.66, + "learning_rate": 0.0005, + "loss": 1.048, + "step": 1524 + }, + { + "epoch": 1.66, + "learning_rate": 0.0005, + "loss": 1.0404, + "step": 1525 + }, + { + "epoch": 1.66, + "learning_rate": 0.0005, + "loss": 1.0575, + "step": 1526 + }, + { + "epoch": 1.66, + "learning_rate": 0.0005, + "loss": 1.0894, + "step": 1527 + }, + { + "epoch": 1.66, + "learning_rate": 0.0005, + "loss": 1.0451, + "step": 1528 + }, + { + "epoch": 1.66, + "learning_rate": 0.0005, + "loss": 1.0604, + "step": 1529 + }, + { + "epoch": 1.66, + "learning_rate": 0.0005, + "loss": 1.0238, + "step": 1530 + }, + { + "epoch": 1.67, + "learning_rate": 0.0005, + "loss": 1.0596, + "step": 1531 + }, + { + "epoch": 1.67, + "learning_rate": 0.0005, + "loss": 1.0614, + "step": 1532 + }, + { + "epoch": 1.67, + "learning_rate": 0.0005, + "loss": 1.0824, + "step": 1533 + }, + { + "epoch": 1.67, + "learning_rate": 0.0005, + "loss": 1.011, + "step": 1534 + }, + { + "epoch": 1.67, + "learning_rate": 0.0005, + "loss": 1.0734, + "step": 1535 + }, + { + "epoch": 1.67, + "learning_rate": 0.0005, + "loss": 1.0084, + "step": 1536 + }, + { + "epoch": 1.67, + "learning_rate": 0.0005, + "loss": 1.0532, + "step": 1537 + }, + { + "epoch": 1.67, + "learning_rate": 0.0005, + "loss": 1.0816, + "step": 1538 + }, + { + "epoch": 1.67, + "learning_rate": 0.0005, + "loss": 0.9924, + "step": 1539 + }, + { + "epoch": 1.68, + "learning_rate": 0.0005, + "loss": 1.0984, + "step": 1540 + }, + { + "epoch": 1.68, + "learning_rate": 0.0005, + "loss": 1.0629, + "step": 1541 + }, + { + "epoch": 1.68, + "learning_rate": 0.0005, + "loss": 1.0967, + "step": 1542 + }, + { + "epoch": 1.68, + "learning_rate": 0.0005, + "loss": 1.0098, + "step": 1543 + }, + { + "epoch": 1.68, + "learning_rate": 0.0005, + "loss": 1.0524, + "step": 1544 + }, + { + "epoch": 1.68, + "learning_rate": 0.0005, + "loss": 1.0525, + "step": 1545 + }, + { + "epoch": 1.68, + "learning_rate": 0.0005, + "loss": 1.1059, + "step": 1546 + }, + { + "epoch": 1.68, + "learning_rate": 0.0005, + "loss": 1.0487, + "step": 1547 + }, + { + "epoch": 1.68, + "learning_rate": 0.0005, + "loss": 1.0287, + "step": 1548 + }, + { + "epoch": 1.69, + "learning_rate": 0.0005, + "loss": 1.0832, + "step": 1549 + }, + { + "epoch": 1.69, + "learning_rate": 0.0005, + "loss": 1.0174, + "step": 1550 + }, + { + "epoch": 1.69, + "learning_rate": 0.0005, + "loss": 1.0217, + "step": 1551 + }, + { + "epoch": 1.69, + "learning_rate": 0.0005, + "loss": 1.0234, + "step": 1552 + }, + { + "epoch": 1.69, + "learning_rate": 0.0005, + "loss": 1.0928, + "step": 1553 + }, + { + "epoch": 1.69, + "learning_rate": 0.0005, + "loss": 1.0952, + "step": 1554 + }, + { + "epoch": 1.69, + "learning_rate": 0.0005, + "loss": 1.0764, + "step": 1555 + }, + { + "epoch": 1.69, + "learning_rate": 0.0005, + "loss": 1.0291, + "step": 1556 + }, + { + "epoch": 1.69, + "learning_rate": 0.0005, + "loss": 1.0315, + "step": 1557 + }, + { + "epoch": 1.7, + "learning_rate": 0.0005, + "loss": 1.0314, + "step": 1558 + }, + { + "epoch": 1.7, + "learning_rate": 0.0005, + "loss": 1.021, + "step": 1559 + }, + { + "epoch": 1.7, + "learning_rate": 0.0005, + "loss": 1.0709, + "step": 1560 + }, + { + "epoch": 1.7, + "learning_rate": 0.0005, + "loss": 1.0619, + "step": 1561 + }, + { + "epoch": 1.7, + "learning_rate": 0.0005, + "loss": 1.0933, + "step": 1562 + }, + { + "epoch": 1.7, + "learning_rate": 0.0005, + "loss": 1.0762, + "step": 1563 + }, + { + "epoch": 1.7, + "learning_rate": 0.0005, + "loss": 1.0545, + "step": 1564 + }, + { + "epoch": 1.7, + "learning_rate": 0.0005, + "loss": 1.0614, + "step": 1565 + }, + { + "epoch": 1.7, + "learning_rate": 0.0005, + "loss": 1.0676, + "step": 1566 + }, + { + "epoch": 1.71, + "learning_rate": 0.0005, + "loss": 1.0421, + "step": 1567 + }, + { + "epoch": 1.71, + "learning_rate": 0.0005, + "loss": 1.0502, + "step": 1568 + }, + { + "epoch": 1.71, + "learning_rate": 0.0005, + "loss": 1.0742, + "step": 1569 + }, + { + "epoch": 1.71, + "learning_rate": 0.0005, + "loss": 1.1046, + "step": 1570 + }, + { + "epoch": 1.71, + "learning_rate": 0.0005, + "loss": 1.0883, + "step": 1571 + }, + { + "epoch": 1.71, + "learning_rate": 0.0005, + "loss": 1.0304, + "step": 1572 + }, + { + "epoch": 1.71, + "learning_rate": 0.0005, + "loss": 1.0489, + "step": 1573 + }, + { + "epoch": 1.71, + "learning_rate": 0.0005, + "loss": 1.088, + "step": 1574 + }, + { + "epoch": 1.71, + "learning_rate": 0.0005, + "loss": 1.0506, + "step": 1575 + }, + { + "epoch": 1.71, + "learning_rate": 0.0005, + "loss": 1.0782, + "step": 1576 + }, + { + "epoch": 1.72, + "learning_rate": 0.0005, + "loss": 1.0745, + "step": 1577 + }, + { + "epoch": 1.72, + "learning_rate": 0.0005, + "loss": 1.0358, + "step": 1578 + }, + { + "epoch": 1.72, + "learning_rate": 0.0005, + "loss": 1.0887, + "step": 1579 + }, + { + "epoch": 1.72, + "learning_rate": 0.0005, + "loss": 1.0237, + "step": 1580 + }, + { + "epoch": 1.72, + "learning_rate": 0.0005, + "loss": 1.0479, + "step": 1581 + }, + { + "epoch": 1.72, + "learning_rate": 0.0005, + "loss": 1.0784, + "step": 1582 + }, + { + "epoch": 1.72, + "learning_rate": 0.0005, + "loss": 1.051, + "step": 1583 + }, + { + "epoch": 1.72, + "learning_rate": 0.0005, + "loss": 1.0446, + "step": 1584 + }, + { + "epoch": 1.72, + "learning_rate": 0.0005, + "loss": 1.0749, + "step": 1585 + }, + { + "epoch": 1.73, + "learning_rate": 0.0005, + "loss": 1.0671, + "step": 1586 + }, + { + "epoch": 1.73, + "learning_rate": 0.0005, + "loss": 1.0831, + "step": 1587 + }, + { + "epoch": 1.73, + "learning_rate": 0.0005, + "loss": 1.0737, + "step": 1588 + }, + { + "epoch": 1.73, + "learning_rate": 0.0005, + "loss": 1.0625, + "step": 1589 + }, + { + "epoch": 1.73, + "learning_rate": 0.0005, + "loss": 1.0901, + "step": 1590 + }, + { + "epoch": 1.73, + "learning_rate": 0.0005, + "loss": 1.0326, + "step": 1591 + }, + { + "epoch": 1.73, + "learning_rate": 0.0005, + "loss": 1.0958, + "step": 1592 + }, + { + "epoch": 1.73, + "learning_rate": 0.0005, + "loss": 1.0333, + "step": 1593 + }, + { + "epoch": 1.73, + "learning_rate": 0.0005, + "loss": 1.0099, + "step": 1594 + }, + { + "epoch": 1.74, + "learning_rate": 0.0005, + "loss": 1.0704, + "step": 1595 + }, + { + "epoch": 1.74, + "learning_rate": 0.0005, + "loss": 1.0212, + "step": 1596 + }, + { + "epoch": 1.74, + "learning_rate": 0.0005, + "loss": 1.0552, + "step": 1597 + }, + { + "epoch": 1.74, + "learning_rate": 0.0005, + "loss": 1.0632, + "step": 1598 + }, + { + "epoch": 1.74, + "learning_rate": 0.0005, + "loss": 1.0507, + "step": 1599 + }, + { + "epoch": 1.74, + "learning_rate": 0.0005, + "loss": 1.0833, + "step": 1600 + }, + { + "epoch": 1.74, + "learning_rate": 0.0005, + "loss": 1.053, + "step": 1601 + }, + { + "epoch": 1.74, + "learning_rate": 0.0005, + "loss": 1.089, + "step": 1602 + }, + { + "epoch": 1.74, + "learning_rate": 0.0005, + "loss": 1.0819, + "step": 1603 + }, + { + "epoch": 1.75, + "learning_rate": 0.0005, + "loss": 1.0626, + "step": 1604 + }, + { + "epoch": 1.75, + "learning_rate": 0.0005, + "loss": 1.0792, + "step": 1605 + }, + { + "epoch": 1.75, + "learning_rate": 0.0005, + "loss": 1.0479, + "step": 1606 + }, + { + "epoch": 1.75, + "learning_rate": 0.0005, + "loss": 1.0552, + "step": 1607 + }, + { + "epoch": 1.75, + "learning_rate": 0.0005, + "loss": 1.0918, + "step": 1608 + }, + { + "epoch": 1.75, + "learning_rate": 0.0005, + "loss": 1.0097, + "step": 1609 + }, + { + "epoch": 1.75, + "learning_rate": 0.0005, + "loss": 1.063, + "step": 1610 + }, + { + "epoch": 1.75, + "learning_rate": 0.0005, + "loss": 1.0025, + "step": 1611 + }, + { + "epoch": 1.75, + "learning_rate": 0.0005, + "loss": 1.0605, + "step": 1612 + }, + { + "epoch": 1.76, + "learning_rate": 0.0005, + "loss": 1.0355, + "step": 1613 + }, + { + "epoch": 1.76, + "learning_rate": 0.0005, + "loss": 1.0145, + "step": 1614 + }, + { + "epoch": 1.76, + "learning_rate": 0.0005, + "loss": 1.0332, + "step": 1615 + }, + { + "epoch": 1.76, + "learning_rate": 0.0005, + "loss": 1.078, + "step": 1616 + }, + { + "epoch": 1.76, + "learning_rate": 0.0005, + "loss": 1.075, + "step": 1617 + }, + { + "epoch": 1.76, + "learning_rate": 0.0005, + "loss": 1.0411, + "step": 1618 + }, + { + "epoch": 1.76, + "learning_rate": 0.0005, + "loss": 1.0593, + "step": 1619 + }, + { + "epoch": 1.76, + "learning_rate": 0.0005, + "loss": 1.061, + "step": 1620 + }, + { + "epoch": 1.76, + "learning_rate": 0.0005, + "loss": 1.0965, + "step": 1621 + }, + { + "epoch": 1.76, + "learning_rate": 0.0005, + "loss": 1.0603, + "step": 1622 + }, + { + "epoch": 1.77, + "learning_rate": 0.0005, + "loss": 1.0635, + "step": 1623 + }, + { + "epoch": 1.77, + "learning_rate": 0.0005, + "loss": 1.064, + "step": 1624 + }, + { + "epoch": 1.77, + "learning_rate": 0.0005, + "loss": 1.0716, + "step": 1625 + }, + { + "epoch": 1.77, + "learning_rate": 0.0005, + "loss": 1.019, + "step": 1626 + }, + { + "epoch": 1.77, + "learning_rate": 0.0005, + "loss": 1.0591, + "step": 1627 + }, + { + "epoch": 1.77, + "learning_rate": 0.0005, + "loss": 1.0797, + "step": 1628 + }, + { + "epoch": 1.77, + "learning_rate": 0.0005, + "loss": 1.0571, + "step": 1629 + }, + { + "epoch": 1.77, + "learning_rate": 0.0005, + "loss": 1.0656, + "step": 1630 + }, + { + "epoch": 1.77, + "learning_rate": 0.0005, + "loss": 1.0441, + "step": 1631 + }, + { + "epoch": 1.78, + "learning_rate": 0.0005, + "loss": 1.0459, + "step": 1632 + }, + { + "epoch": 1.78, + "learning_rate": 0.0005, + "loss": 1.079, + "step": 1633 + }, + { + "epoch": 1.78, + "learning_rate": 0.0005, + "loss": 1.0897, + "step": 1634 + }, + { + "epoch": 1.78, + "learning_rate": 0.0005, + "loss": 1.0385, + "step": 1635 + }, + { + "epoch": 1.78, + "learning_rate": 0.0005, + "loss": 1.0645, + "step": 1636 + }, + { + "epoch": 1.78, + "learning_rate": 0.0005, + "loss": 1.1057, + "step": 1637 + }, + { + "epoch": 1.78, + "learning_rate": 0.0005, + "loss": 1.0806, + "step": 1638 + }, + { + "epoch": 1.78, + "learning_rate": 0.0005, + "loss": 1.0689, + "step": 1639 + }, + { + "epoch": 1.78, + "learning_rate": 0.0005, + "loss": 1.0459, + "step": 1640 + }, + { + "epoch": 1.79, + "learning_rate": 0.0005, + "loss": 1.0513, + "step": 1641 + }, + { + "epoch": 1.79, + "learning_rate": 0.0005, + "loss": 1.0603, + "step": 1642 + }, + { + "epoch": 1.79, + "learning_rate": 0.0005, + "loss": 1.0314, + "step": 1643 + }, + { + "epoch": 1.79, + "learning_rate": 0.0005, + "loss": 1.1045, + "step": 1644 + }, + { + "epoch": 1.79, + "learning_rate": 0.0005, + "loss": 1.0751, + "step": 1645 + }, + { + "epoch": 1.79, + "learning_rate": 0.0005, + "loss": 1.0261, + "step": 1646 + }, + { + "epoch": 1.79, + "learning_rate": 0.0005, + "loss": 1.0526, + "step": 1647 + }, + { + "epoch": 1.79, + "learning_rate": 0.0005, + "loss": 1.0492, + "step": 1648 + }, + { + "epoch": 1.79, + "learning_rate": 0.0005, + "loss": 1.0717, + "step": 1649 + }, + { + "epoch": 1.8, + "learning_rate": 0.0005, + "loss": 1.044, + "step": 1650 + }, + { + "epoch": 1.8, + "learning_rate": 0.0005, + "loss": 1.0619, + "step": 1651 + }, + { + "epoch": 1.8, + "learning_rate": 0.0005, + "loss": 1.0732, + "step": 1652 + }, + { + "epoch": 1.8, + "learning_rate": 0.0005, + "loss": 1.0361, + "step": 1653 + }, + { + "epoch": 1.8, + "learning_rate": 0.0005, + "loss": 1.0437, + "step": 1654 + }, + { + "epoch": 1.8, + "learning_rate": 0.0005, + "loss": 1.0484, + "step": 1655 + }, + { + "epoch": 1.8, + "learning_rate": 0.0005, + "loss": 1.0402, + "step": 1656 + }, + { + "epoch": 1.8, + "learning_rate": 0.0005, + "loss": 1.0561, + "step": 1657 + }, + { + "epoch": 1.8, + "learning_rate": 0.0005, + "loss": 1.0439, + "step": 1658 + }, + { + "epoch": 1.81, + "learning_rate": 0.0005, + "loss": 1.053, + "step": 1659 + }, + { + "epoch": 1.81, + "learning_rate": 0.0005, + "loss": 1.0834, + "step": 1660 + }, + { + "epoch": 1.81, + "learning_rate": 0.0005, + "loss": 1.0573, + "step": 1661 + }, + { + "epoch": 1.81, + "learning_rate": 0.0005, + "loss": 1.0059, + "step": 1662 + }, + { + "epoch": 1.81, + "learning_rate": 0.0005, + "loss": 1.0951, + "step": 1663 + }, + { + "epoch": 1.81, + "learning_rate": 0.0005, + "loss": 1.0494, + "step": 1664 + }, + { + "epoch": 1.81, + "learning_rate": 0.0005, + "loss": 0.9884, + "step": 1665 + }, + { + "epoch": 1.81, + "learning_rate": 0.0005, + "loss": 1.0532, + "step": 1666 + }, + { + "epoch": 1.81, + "learning_rate": 0.0005, + "loss": 1.041, + "step": 1667 + }, + { + "epoch": 1.82, + "learning_rate": 0.0005, + "loss": 1.0319, + "step": 1668 + }, + { + "epoch": 1.82, + "learning_rate": 0.0005, + "loss": 1.0076, + "step": 1669 + }, + { + "epoch": 1.82, + "learning_rate": 0.0005, + "loss": 1.0338, + "step": 1670 + }, + { + "epoch": 1.82, + "learning_rate": 0.0005, + "loss": 1.0429, + "step": 1671 + }, + { + "epoch": 1.82, + "learning_rate": 0.0005, + "loss": 1.0339, + "step": 1672 + }, + { + "epoch": 1.82, + "learning_rate": 0.0005, + "loss": 1.0403, + "step": 1673 + }, + { + "epoch": 1.82, + "learning_rate": 0.0005, + "loss": 1.0692, + "step": 1674 + }, + { + "epoch": 1.82, + "learning_rate": 0.0005, + "loss": 1.0484, + "step": 1675 + }, + { + "epoch": 1.82, + "learning_rate": 0.0005, + "loss": 1.0219, + "step": 1676 + }, + { + "epoch": 1.82, + "learning_rate": 0.0005, + "loss": 1.0377, + "step": 1677 + }, + { + "epoch": 1.83, + "learning_rate": 0.0005, + "loss": 1.0893, + "step": 1678 + }, + { + "epoch": 1.83, + "learning_rate": 0.0005, + "loss": 1.0305, + "step": 1679 + }, + { + "epoch": 1.83, + "learning_rate": 0.0005, + "loss": 1.0356, + "step": 1680 + }, + { + "epoch": 1.83, + "learning_rate": 0.0005, + "loss": 1.0289, + "step": 1681 + }, + { + "epoch": 1.83, + "learning_rate": 0.0005, + "loss": 1.0474, + "step": 1682 + }, + { + "epoch": 1.83, + "learning_rate": 0.0005, + "loss": 1.0609, + "step": 1683 + }, + { + "epoch": 1.83, + "learning_rate": 0.0005, + "loss": 1.0321, + "step": 1684 + }, + { + "epoch": 1.83, + "learning_rate": 0.0005, + "loss": 1.0816, + "step": 1685 + }, + { + "epoch": 1.83, + "learning_rate": 0.0005, + "loss": 1.0923, + "step": 1686 + }, + { + "epoch": 1.84, + "learning_rate": 0.0005, + "loss": 1.0432, + "step": 1687 + }, + { + "epoch": 1.84, + "learning_rate": 0.0005, + "loss": 1.0323, + "step": 1688 + }, + { + "epoch": 1.84, + "learning_rate": 0.0005, + "loss": 1.061, + "step": 1689 + }, + { + "epoch": 1.84, + "learning_rate": 0.0005, + "loss": 1.0117, + "step": 1690 + }, + { + "epoch": 1.84, + "learning_rate": 0.0005, + "loss": 1.0792, + "step": 1691 + }, + { + "epoch": 1.84, + "learning_rate": 0.0005, + "loss": 1.077, + "step": 1692 + }, + { + "epoch": 1.84, + "learning_rate": 0.0005, + "loss": 1.0946, + "step": 1693 + }, + { + "epoch": 1.84, + "learning_rate": 0.0005, + "loss": 1.0646, + "step": 1694 + }, + { + "epoch": 1.84, + "learning_rate": 0.0005, + "loss": 1.0308, + "step": 1695 + }, + { + "epoch": 1.85, + "learning_rate": 0.0005, + "loss": 1.0905, + "step": 1696 + }, + { + "epoch": 1.85, + "learning_rate": 0.0005, + "loss": 1.0325, + "step": 1697 + }, + { + "epoch": 1.85, + "learning_rate": 0.0005, + "loss": 1.0549, + "step": 1698 + }, + { + "epoch": 1.85, + "learning_rate": 0.0005, + "loss": 1.0172, + "step": 1699 + }, + { + "epoch": 1.85, + "learning_rate": 0.0005, + "loss": 1.0317, + "step": 1700 + }, + { + "epoch": 1.85, + "learning_rate": 0.0005, + "loss": 1.0851, + "step": 1701 + }, + { + "epoch": 1.85, + "learning_rate": 0.0005, + "loss": 1.0508, + "step": 1702 + }, + { + "epoch": 1.85, + "learning_rate": 0.0005, + "loss": 1.0233, + "step": 1703 + }, + { + "epoch": 1.85, + "learning_rate": 0.0005, + "loss": 1.0274, + "step": 1704 + }, + { + "epoch": 1.86, + "learning_rate": 0.0005, + "loss": 1.0875, + "step": 1705 + }, + { + "epoch": 1.86, + "learning_rate": 0.0005, + "loss": 1.0269, + "step": 1706 + }, + { + "epoch": 1.86, + "learning_rate": 0.0005, + "loss": 1.0538, + "step": 1707 + }, + { + "epoch": 1.86, + "learning_rate": 0.0005, + "loss": 1.0355, + "step": 1708 + }, + { + "epoch": 1.86, + "learning_rate": 0.0005, + "loss": 1.0423, + "step": 1709 + }, + { + "epoch": 1.86, + "learning_rate": 0.0005, + "loss": 1.0925, + "step": 1710 + }, + { + "epoch": 1.86, + "learning_rate": 0.0005, + "loss": 1.0287, + "step": 1711 + }, + { + "epoch": 1.86, + "learning_rate": 0.0005, + "loss": 1.0398, + "step": 1712 + }, + { + "epoch": 1.86, + "learning_rate": 0.0005, + "loss": 1.0138, + "step": 1713 + }, + { + "epoch": 1.87, + "learning_rate": 0.0005, + "loss": 1.0737, + "step": 1714 + }, + { + "epoch": 1.87, + "learning_rate": 0.0005, + "loss": 1.0773, + "step": 1715 + }, + { + "epoch": 1.87, + "learning_rate": 0.0005, + "loss": 1.085, + "step": 1716 + }, + { + "epoch": 1.87, + "learning_rate": 0.0005, + "loss": 1.0689, + "step": 1717 + }, + { + "epoch": 1.87, + "learning_rate": 0.0005, + "loss": 1.0811, + "step": 1718 + }, + { + "epoch": 1.87, + "learning_rate": 0.0005, + "loss": 1.0546, + "step": 1719 + }, + { + "epoch": 1.87, + "learning_rate": 0.0005, + "loss": 1.0501, + "step": 1720 + }, + { + "epoch": 1.87, + "learning_rate": 0.0005, + "loss": 1.1132, + "step": 1721 + }, + { + "epoch": 1.87, + "learning_rate": 0.0005, + "loss": 1.0998, + "step": 1722 + }, + { + "epoch": 1.87, + "learning_rate": 0.0005, + "loss": 1.0375, + "step": 1723 + }, + { + "epoch": 1.88, + "learning_rate": 0.0005, + "loss": 1.065, + "step": 1724 + }, + { + "epoch": 1.88, + "learning_rate": 0.0005, + "loss": 1.0681, + "step": 1725 + }, + { + "epoch": 1.88, + "learning_rate": 0.0005, + "loss": 1.0238, + "step": 1726 + }, + { + "epoch": 1.88, + "learning_rate": 0.0005, + "loss": 1.0679, + "step": 1727 + }, + { + "epoch": 1.88, + "learning_rate": 0.0005, + "loss": 1.0728, + "step": 1728 + }, + { + "epoch": 1.88, + "learning_rate": 0.0005, + "loss": 1.0346, + "step": 1729 + }, + { + "epoch": 1.88, + "learning_rate": 0.0005, + "loss": 1.011, + "step": 1730 + }, + { + "epoch": 1.88, + "learning_rate": 0.0005, + "loss": 1.0572, + "step": 1731 + }, + { + "epoch": 1.88, + "learning_rate": 0.0005, + "loss": 1.0437, + "step": 1732 + }, + { + "epoch": 1.89, + "learning_rate": 0.0005, + "loss": 1.0685, + "step": 1733 + }, + { + "epoch": 1.89, + "learning_rate": 0.0005, + "loss": 1.0509, + "step": 1734 + }, + { + "epoch": 1.89, + "learning_rate": 0.0005, + "loss": 1.0367, + "step": 1735 + }, + { + "epoch": 1.89, + "learning_rate": 0.0005, + "loss": 1.0474, + "step": 1736 + }, + { + "epoch": 1.89, + "learning_rate": 0.0005, + "loss": 1.0965, + "step": 1737 + }, + { + "epoch": 1.89, + "learning_rate": 0.0005, + "loss": 1.0539, + "step": 1738 + }, + { + "epoch": 1.89, + "learning_rate": 0.0005, + "loss": 1.0411, + "step": 1739 + }, + { + "epoch": 1.89, + "learning_rate": 0.0005, + "loss": 1.0958, + "step": 1740 + }, + { + "epoch": 1.89, + "learning_rate": 0.0005, + "loss": 1.0571, + "step": 1741 + }, + { + "epoch": 1.9, + "learning_rate": 0.0005, + "loss": 1.0606, + "step": 1742 + }, + { + "epoch": 1.9, + "learning_rate": 0.0005, + "loss": 0.9869, + "step": 1743 + }, + { + "epoch": 1.9, + "learning_rate": 0.0005, + "loss": 0.982, + "step": 1744 + }, + { + "epoch": 1.9, + "learning_rate": 0.0005, + "loss": 1.0301, + "step": 1745 + }, + { + "epoch": 1.9, + "learning_rate": 0.0005, + "loss": 1.0464, + "step": 1746 + }, + { + "epoch": 1.9, + "learning_rate": 0.0005, + "loss": 1.0359, + "step": 1747 + }, + { + "epoch": 1.9, + "learning_rate": 0.0005, + "loss": 1.0488, + "step": 1748 + }, + { + "epoch": 1.9, + "learning_rate": 0.0005, + "loss": 1.0457, + "step": 1749 + }, + { + "epoch": 1.9, + "learning_rate": 0.0005, + "loss": 1.0247, + "step": 1750 + }, + { + "epoch": 1.91, + "learning_rate": 0.0005, + "loss": 1.0865, + "step": 1751 + }, + { + "epoch": 1.91, + "learning_rate": 0.0005, + "loss": 1.0731, + "step": 1752 + }, + { + "epoch": 1.91, + "learning_rate": 0.0005, + "loss": 1.0509, + "step": 1753 + }, + { + "epoch": 1.91, + "learning_rate": 0.0005, + "loss": 1.0403, + "step": 1754 + }, + { + "epoch": 1.91, + "learning_rate": 0.0005, + "loss": 1.0599, + "step": 1755 + }, + { + "epoch": 1.91, + "learning_rate": 0.0005, + "loss": 1.0093, + "step": 1756 + }, + { + "epoch": 1.91, + "learning_rate": 0.0005, + "loss": 1.0675, + "step": 1757 + }, + { + "epoch": 1.91, + "learning_rate": 0.0005, + "loss": 1.0352, + "step": 1758 + }, + { + "epoch": 1.91, + "learning_rate": 0.0005, + "loss": 1.0759, + "step": 1759 + }, + { + "epoch": 1.92, + "learning_rate": 0.0005, + "loss": 1.0877, + "step": 1760 + }, + { + "epoch": 1.92, + "learning_rate": 0.0005, + "loss": 1.0704, + "step": 1761 + }, + { + "epoch": 1.92, + "learning_rate": 0.0005, + "loss": 1.0185, + "step": 1762 + }, + { + "epoch": 1.92, + "learning_rate": 0.0005, + "loss": 1.0642, + "step": 1763 + }, + { + "epoch": 1.92, + "learning_rate": 0.0005, + "loss": 1.0451, + "step": 1764 + }, + { + "epoch": 1.92, + "learning_rate": 0.0005, + "loss": 1.0997, + "step": 1765 + }, + { + "epoch": 1.92, + "learning_rate": 0.0005, + "loss": 1.1098, + "step": 1766 + }, + { + "epoch": 1.92, + "learning_rate": 0.0005, + "loss": 1.0718, + "step": 1767 + }, + { + "epoch": 1.92, + "learning_rate": 0.0005, + "loss": 1.0624, + "step": 1768 + }, + { + "epoch": 1.92, + "learning_rate": 0.0005, + "loss": 1.0381, + "step": 1769 + }, + { + "epoch": 1.93, + "learning_rate": 0.0005, + "loss": 1.0529, + "step": 1770 + }, + { + "epoch": 1.93, + "learning_rate": 0.0005, + "loss": 1.0982, + "step": 1771 + }, + { + "epoch": 1.93, + "learning_rate": 0.0005, + "loss": 1.0902, + "step": 1772 + }, + { + "epoch": 1.93, + "learning_rate": 0.0005, + "loss": 1.0767, + "step": 1773 + }, + { + "epoch": 1.93, + "learning_rate": 0.0005, + "loss": 1.0722, + "step": 1774 + }, + { + "epoch": 1.93, + "learning_rate": 0.0005, + "loss": 1.0495, + "step": 1775 + }, + { + "epoch": 1.93, + "learning_rate": 0.0005, + "loss": 1.0685, + "step": 1776 + }, + { + "epoch": 1.93, + "learning_rate": 0.0005, + "loss": 1.0242, + "step": 1777 + }, + { + "epoch": 1.93, + "learning_rate": 0.0005, + "loss": 1.0938, + "step": 1778 + }, + { + "epoch": 1.94, + "learning_rate": 0.0005, + "loss": 1.0219, + "step": 1779 + }, + { + "epoch": 1.94, + "learning_rate": 0.0005, + "loss": 1.0301, + "step": 1780 + }, + { + "epoch": 1.94, + "learning_rate": 0.0005, + "loss": 1.0291, + "step": 1781 + }, + { + "epoch": 1.94, + "learning_rate": 0.0005, + "loss": 1.0181, + "step": 1782 + }, + { + "epoch": 1.94, + "learning_rate": 0.0005, + "loss": 1.0328, + "step": 1783 + }, + { + "epoch": 1.94, + "learning_rate": 0.0005, + "loss": 1.062, + "step": 1784 + }, + { + "epoch": 1.94, + "learning_rate": 0.0005, + "loss": 1.0154, + "step": 1785 + }, + { + "epoch": 1.94, + "learning_rate": 0.0005, + "loss": 1.0713, + "step": 1786 + }, + { + "epoch": 1.94, + "learning_rate": 0.0005, + "loss": 1.0366, + "step": 1787 + }, + { + "epoch": 1.95, + "learning_rate": 0.0005, + "loss": 1.082, + "step": 1788 + }, + { + "epoch": 1.95, + "learning_rate": 0.0005, + "loss": 1.0558, + "step": 1789 + }, + { + "epoch": 1.95, + "learning_rate": 0.0005, + "loss": 1.0296, + "step": 1790 + }, + { + "epoch": 1.95, + "learning_rate": 0.0005, + "loss": 1.1335, + "step": 1791 + }, + { + "epoch": 1.95, + "learning_rate": 0.0005, + "loss": 1.0461, + "step": 1792 + }, + { + "epoch": 1.95, + "learning_rate": 0.0005, + "loss": 1.0634, + "step": 1793 + }, + { + "epoch": 1.95, + "learning_rate": 0.0005, + "loss": 1.0641, + "step": 1794 + }, + { + "epoch": 1.95, + "learning_rate": 0.0005, + "loss": 1.0258, + "step": 1795 + }, + { + "epoch": 1.95, + "learning_rate": 0.0005, + "loss": 1.0613, + "step": 1796 + }, + { + "epoch": 1.96, + "learning_rate": 0.0005, + "loss": 1.0324, + "step": 1797 + }, + { + "epoch": 1.96, + "learning_rate": 0.0005, + "loss": 1.0448, + "step": 1798 + }, + { + "epoch": 1.96, + "learning_rate": 0.0005, + "loss": 1.0015, + "step": 1799 + }, + { + "epoch": 1.96, + "learning_rate": 0.0005, + "loss": 1.0627, + "step": 1800 + }, + { + "epoch": 1.96, + "learning_rate": 0.0005, + "loss": 1.029, + "step": 1801 + }, + { + "epoch": 1.96, + "learning_rate": 0.0005, + "loss": 1.0979, + "step": 1802 + }, + { + "epoch": 1.96, + "learning_rate": 0.0005, + "loss": 1.073, + "step": 1803 + }, + { + "epoch": 1.96, + "learning_rate": 0.0005, + "loss": 1.0498, + "step": 1804 + }, + { + "epoch": 1.96, + "learning_rate": 0.0005, + "loss": 1.045, + "step": 1805 + }, + { + "epoch": 1.97, + "learning_rate": 0.0005, + "loss": 1.0347, + "step": 1806 + }, + { + "epoch": 1.97, + "learning_rate": 0.0005, + "loss": 1.0259, + "step": 1807 + }, + { + "epoch": 1.97, + "learning_rate": 0.0005, + "loss": 1.0202, + "step": 1808 + }, + { + "epoch": 1.97, + "learning_rate": 0.0005, + "loss": 1.0513, + "step": 1809 + }, + { + "epoch": 1.97, + "learning_rate": 0.0005, + "loss": 1.0402, + "step": 1810 + }, + { + "epoch": 1.97, + "learning_rate": 0.0005, + "loss": 1.0973, + "step": 1811 + }, + { + "epoch": 1.97, + "learning_rate": 0.0005, + "loss": 1.08, + "step": 1812 + }, + { + "epoch": 1.97, + "learning_rate": 0.0005, + "loss": 1.0454, + "step": 1813 + }, + { + "epoch": 1.97, + "learning_rate": 0.0005, + "loss": 1.0386, + "step": 1814 + }, + { + "epoch": 1.97, + "learning_rate": 0.0005, + "loss": 1.0853, + "step": 1815 + }, + { + "epoch": 1.98, + "learning_rate": 0.0005, + "loss": 1.0519, + "step": 1816 + }, + { + "epoch": 1.98, + "learning_rate": 0.0005, + "loss": 1.0334, + "step": 1817 + }, + { + "epoch": 1.98, + "learning_rate": 0.0005, + "loss": 1.0579, + "step": 1818 + }, + { + "epoch": 1.98, + "learning_rate": 0.0005, + "loss": 1.0384, + "step": 1819 + }, + { + "epoch": 1.98, + "learning_rate": 0.0005, + "loss": 1.0277, + "step": 1820 + }, + { + "epoch": 1.98, + "learning_rate": 0.0005, + "loss": 1.0007, + "step": 1821 + }, + { + "epoch": 1.98, + "learning_rate": 0.0005, + "loss": 1.0565, + "step": 1822 + }, + { + "epoch": 1.98, + "learning_rate": 0.0005, + "loss": 1.0234, + "step": 1823 + }, + { + "epoch": 1.98, + "learning_rate": 0.0005, + "loss": 1.063, + "step": 1824 + }, + { + "epoch": 1.99, + "learning_rate": 0.0005, + "loss": 1.0622, + "step": 1825 + }, + { + "epoch": 1.99, + "learning_rate": 0.0005, + "loss": 1.0646, + "step": 1826 + }, + { + "epoch": 1.99, + "learning_rate": 0.0005, + "loss": 1.0502, + "step": 1827 + }, + { + "epoch": 1.99, + "learning_rate": 0.0005, + "loss": 1.0666, + "step": 1828 + }, + { + "epoch": 1.99, + "learning_rate": 0.0005, + "loss": 1.1151, + "step": 1829 + }, + { + "epoch": 1.99, + "learning_rate": 0.0005, + "loss": 1.0488, + "step": 1830 + }, + { + "epoch": 1.99, + "learning_rate": 0.0005, + "loss": 1.0301, + "step": 1831 + }, + { + "epoch": 1.99, + "learning_rate": 0.0005, + "loss": 1.0923, + "step": 1832 + }, + { + "epoch": 1.99, + "learning_rate": 0.0005, + "loss": 1.0564, + "step": 1833 + }, + { + "epoch": 2.0, + "learning_rate": 0.0005, + "loss": 1.0337, + "step": 1834 + }, + { + "epoch": 2.0, + "learning_rate": 0.0005, + "loss": 1.0216, + "step": 1835 + }, + { + "epoch": 2.0, + "learning_rate": 0.0005, + "loss": 1.0522, + "step": 1836 + }, + { + "epoch": 2.0, + "learning_rate": 0.0005, + "loss": 1.0773, + "step": 1837 + }, + { + "epoch": 2.0, + "learning_rate": 0.0005, + "loss": 1.0832, + "step": 1838 + } + ], + "logging_steps": 1, + "max_steps": 6000, + "num_train_epochs": 7, + "save_steps": 500, + "total_flos": 4.642446064025764e+17, + "trial_name": null, + "trial_params": null +}